Files
pikasTech-unidesk/scripts/src/platform-infra-sub2api-codex/remote-python-sync.ts
T
2026-06-25 16:16:25 +00:00

2994 lines
133 KiB
TypeScript

// SPEC: PJ2026-01060307 控制面模块化 draft-2026-06-25-p0. remote-python-sync module for scripts/src/platform-infra-sub2api-codex.ts.
// Moved mechanically from scripts/src/platform-infra-sub2api-codex.ts:4265-5400 for #903.
import { chmodSync, copyFileSync, existsSync, mkdirSync, readFileSync, statSync, writeFileSync } from "node:fs";
import { homedir } from "node:os";
import { join } from "node:path";
import type { UniDeskConfig } from "../config";
import { rootPath } from "../config";
import type { RenderedCliResult } from "../output";
import { applyPk01CaddyBlock, prepareFrpcSecret, renderFrpcManifest, type PublicServiceExposure, type PublicServiceTarget } from "../platform-infra-public-service";
import { shortSha256Fingerprint } from "../platform-infra-ops-library";
import {
codexPoolSentinelSummary,
codexPoolSentinelRuntimeImage,
readCodexPoolSentinelConfig,
renderCodexPoolSentinelManifest,
type CodexPoolSentinelConfig,
type CodexPoolSentinelProfileSecret,
} from "../platform-infra-sub2api-codex-sentinel";
import { parseEnvFile, readTextFile, redactRepoPath, requiredEnvValue } from "../secrets";
import { runSshCommandCapture, type SshCaptureResult } from "../ssh";
import type { CodexPoolConfig, CodexPoolRuntimeTarget } from "./types";
import { desiredAccountCapacityMap, desiredAccountLoadFactorMap, desiredAccountTempUnschedulableMap, desiredAccountWebSocketsV2ModeMap } from "./accounts";
import { resolvedManualAccountProtections } from "./public-exposure";
import { fieldManager } from "./types";
export function remotePythonScript(mode: "sync" | "validate" | "trace" | "cleanup-probes" | "sentinel-probe", encodedPayload: string, pool: CodexPoolConfig, target: CodexPoolRuntimeTarget): string {
return `
set -u
python3 - <<'PY'
import base64
import json
import re
import secrets
import string
import subprocess
import sys
import time
from datetime import datetime, timezone, timedelta
from urllib.parse import quote
TARGET_ID = ${JSON.stringify(target.id)}
NAMESPACE = ${JSON.stringify(target.namespace)}
SERVICE_NAME = ${JSON.stringify(target.serviceName)}
SERVICE_DNS = ${JSON.stringify(target.serviceDns)}
FIELD_MANAGER = "${fieldManager}"
APP_SECRET_NAME = ${JSON.stringify(target.appSecretName)}
POOL_GROUP_NAME = "${pool.groupName}"
POOL_GROUP_DESCRIPTION = ${JSON.stringify(pool.groupDescription)}
POOL_API_KEY_NAME = "${pool.apiKeyName}"
POOL_API_KEY_SECRET_NAME = "${pool.apiKeySecretName}"
POOL_API_KEY_SECRET_KEY = "${pool.apiKeySecretKey}"
POOL_ADMIN_EMAIL_DEFAULT = ${JSON.stringify(pool.adminEmailDefault)}
MIN_OWNER_BALANCE_USD = ${JSON.stringify(pool.minOwnerBalanceUsd)}
MIN_OWNER_CONCURRENCY = ${JSON.stringify(pool.minOwnerConcurrency)}
MIN_OWNER_CONCURRENCY_SOURCE = ${JSON.stringify(pool.minOwnerConcurrencySource)}
POOL_DEFAULT_ACCOUNT_PRIORITY = ${JSON.stringify(pool.defaultAccountPriority)}
POOL_DEFAULT_ACCOUNT_CAPACITY = ${JSON.stringify(pool.defaultAccountCapacity)}
POOL_DEFAULT_ACCOUNT_LOAD_FACTOR = ${JSON.stringify(pool.defaultAccountLoadFactor)}
RESPONSES_SMOKE_MODEL = ${JSON.stringify(pool.localCodex.responsesSmokeModel)}
EXPECTED_ACCOUNT_CAPACITIES = ${JSON.stringify(desiredAccountCapacityMap(pool))}
EXPECTED_ACCOUNT_LOAD_FACTORS = ${JSON.stringify(desiredAccountLoadFactorMap(pool))}
EXPECTED_ACCOUNT_WS_MODES = json.loads(${JSON.stringify(JSON.stringify(desiredAccountWebSocketsV2ModeMap(pool)))})
EXPECTED_ACCOUNT_TEMP_UNSCHEDULABLE = json.loads(${JSON.stringify(JSON.stringify(desiredAccountTempUnschedulableMap(pool)))})
MANUAL_ACCOUNT_PROTECTIONS = json.loads(${JSON.stringify(JSON.stringify(resolvedManualAccountProtections(pool, target)))})
SENTINEL_CONFIG = json.loads(${JSON.stringify(JSON.stringify(pool.sentinel))})
TARGET_EGRESS_PROXY = json.loads(${JSON.stringify(JSON.stringify(target.egressProxy))})
MODE = "${mode}"
PAYLOAD_B64 = "${encodedPayload}"
def run(cmd, input_bytes=None):
return subprocess.run(cmd, input=input_bytes, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
def text(data, limit=4000):
if isinstance(data, bytes):
data = data.decode("utf-8", errors="replace")
return data[-limit:]
def kubectl(args, input_obj=None):
if isinstance(input_obj, str):
input_bytes = input_obj.encode("utf-8")
else:
input_bytes = input_obj
return run(["kubectl", *args], input_bytes)
def require_kubectl(args, input_obj=None, label="kubectl"):
proc = kubectl(args, input_obj)
if proc.returncode != 0:
raise RuntimeError(f"{label} failed: {text(proc.stderr, 1000)}")
return proc.stdout
def kube_json(args, label):
raw = require_kubectl([*args, "-o", "json"], label=label)
return json.loads(raw.decode("utf-8"))
def decode_secret_value(name, key):
data = kube_json(["-n", NAMESPACE, "get", "secret", name], f"secret/{name}").get("data") or {}
if key not in data:
return None
return base64.b64decode(data[key]).decode("utf-8")
def get_config_value(name, key):
data = kube_json(["-n", NAMESPACE, "get", "configmap", name], f"configmap/{name}").get("data") or {}
value = data.get(key)
return value if isinstance(value, str) and value else None
def select_app_pod():
pods = kube_json(["-n", NAMESPACE, "get", "pods", "-l", "app.kubernetes.io/name=sub2api"], "sub2api pods").get("items") or []
for pod in pods:
status = pod.get("status") or {}
if status.get("phase") != "Running":
continue
statuses = status.get("containerStatuses") or []
if statuses and all(item.get("ready") is True for item in statuses):
return pod["metadata"]["name"]
if pods:
return pods[0]["metadata"]["name"]
raise RuntimeError("sub2api app pod not found")
APP_POD = select_app_pod()
def parse_curl_output(proc):
stdout = proc.stdout.decode("utf-8", errors="replace")
marker = "\\n__HTTP_CODE__:"
pos = stdout.rfind(marker)
if pos < 0:
return {
"ok": False,
"httpStatus": 0,
"json": None,
"body": stdout,
"stderr": text(proc.stderr, 1000),
"transportExitCode": proc.returncode,
}
body = stdout[:pos]
status_text = stdout[pos + len(marker):].strip()
try:
http_status = int(status_text[-3:])
except ValueError:
http_status = 0
try:
parsed = json.loads(body) if body.strip() else None
except json.JSONDecodeError:
parsed = None
return {
"ok": proc.returncode == 0 and 200 <= http_status < 300,
"httpStatus": http_status,
"json": parsed,
"body": body,
"stderr": text(proc.stderr, 1000),
"transportExitCode": proc.returncode,
}
def utc_iso(offset_seconds=0):
return (datetime.now(timezone.utc) + timedelta(seconds=offset_seconds)).strftime("%Y-%m-%dT%H:%M:%SZ")
def normalize_runtime_base_url(value):
if not isinstance(value, str):
return None
value = value.strip().rstrip("/")
return value or None
def empty_to_none(value):
return value if isinstance(value, str) and value else None
def sentinel_quality_gate_enabled():
return (SENTINEL_CONFIG.get("monitor") or {}).get("enabled") is True and (SENTINEL_CONFIG.get("actions") or {}).get("enabled") is True
def account_notes_fingerprint(account):
notes = account.get("notes") if isinstance(account, dict) else None
if not isinstance(notes, str):
return None
match = re.search(r"fingerprint=([A-Za-z0-9_-]+)", notes)
return match.group(1) if match else None
def runtime_account_credentials(account):
credentials = account.get("credentials") if isinstance(account, dict) and isinstance(account.get("credentials"), dict) else {}
return credentials
def runtime_account_extra(account):
extra = account.get("extra") if isinstance(account, dict) and isinstance(account.get("extra"), dict) else {}
return extra
def sentinel_probe_change_reasons(current, profile):
if not isinstance(current, dict) or current.get("id") is None:
return ["created"]
credentials = runtime_account_credentials(current)
extra = runtime_account_extra(current)
expected_base_url = normalize_runtime_base_url(profile.get("baseUrl"))
runtime_base_url = normalize_runtime_base_url(credentials.get("base_url"))
expected_user_agent = empty_to_none(profile.get("upstreamUserAgent"))
runtime_user_agent = empty_to_none(credentials.get("user_agent"))
expected_ws_mode = empty_to_none(profile.get("openaiResponsesWebSocketsV2Mode"))
runtime_ws_mode = empty_to_none(extra.get("openai_apikey_responses_websockets_v2_mode"))
expected_trust_upstream = profile.get("trustUpstream") is True
runtime_trust_upstream = extra.get("unidesk_trust_upstream") is True
expected_protect = profile.get("sentinelProtect") if isinstance(profile.get("sentinelProtect"), dict) else {"enabled": False}
runtime_protect = extra.get("unidesk_sentinel_protect") if isinstance(extra.get("unidesk_sentinel_protect"), dict) else {"enabled": False}
reasons = []
if empty_to_none(extra.get("unidesk_codex_profile")) != profile.get("profile"):
reasons.append("profile")
if runtime_base_url != expected_base_url:
reasons.append("base-url")
if account_notes_fingerprint(current) != profile.get("apiKeyFingerprint"):
reasons.append("api-key-fingerprint")
if runtime_user_agent != expected_user_agent:
reasons.append("upstream-user-agent")
if runtime_ws_mode != expected_ws_mode:
reasons.append("responses-websockets-v2-mode")
if runtime_trust_upstream != expected_trust_upstream:
reasons.append("trust-upstream")
if runtime_protect != expected_protect:
reasons.append("sentinel-protect")
return reasons
def curl_api(method, path, bearer=None, payload=None):
body = b"" if payload is None else json.dumps(payload, separators=(",", ":")).encode("utf-8")
script = r'''
set -eu
method="$1"
url="$2"
token="\${3:-}"
tmp="$(mktemp)"
trap 'rm -f "$tmp"' EXIT
cat > "$tmp"
if [ -n "$token" ]; then
if [ "$method" = "GET" ] && [ ! -s "$tmp" ]; then
curl -sS -w '\\n__HTTP_CODE__:%{http_code}' -X "$method" -H "Authorization: Bearer $token" "$url"
else
curl -sS -w '\\n__HTTP_CODE__:%{http_code}' -X "$method" -H 'Content-Type: application/json' -H "Authorization: Bearer $token" --data-binary @"$tmp" "$url"
fi
else
if [ "$method" = "GET" ] && [ ! -s "$tmp" ]; then
curl -sS -w '\\n__HTTP_CODE__:%{http_code}' -X "$method" "$url"
else
curl -sS -w '\\n__HTTP_CODE__:%{http_code}' -X "$method" -H 'Content-Type: application/json' --data-binary @"$tmp" "$url"
fi
fi
'''
proc = run([
"kubectl", "-n", NAMESPACE, "exec", "-i", APP_POD,
"--", "sh", "-c", script, "sh", method, f"http://127.0.0.1:8080{path}", bearer or "",
], body)
return parse_curl_output(proc)
def envelope_data(parsed):
if isinstance(parsed, dict) and "data" in parsed:
return parsed.get("data")
return parsed
def ensure_success(resp, label):
parsed = resp.get("json")
code = parsed.get("code") if isinstance(parsed, dict) else None
if not resp.get("ok") or (code is not None and code != 0):
message = parsed.get("message") if isinstance(parsed, dict) else text(resp.get("body", ""), 500)
raise RuntimeError(f"{label} failed: http={resp.get('httpStatus')} message={message}")
return envelope_data(parsed)
def ensure_admin_compliance(token):
status_resp = curl_api("GET", "/api/v1/admin/compliance", bearer=token)
if status_resp.get("httpStatus") == 404:
return {
"ok": True,
"action": "not-supported-by-runtime",
"valuesPrinted": False,
}
status = ensure_success(status_resp, "get admin compliance")
if not isinstance(status, dict):
return {
"ok": True,
"action": "status-unstructured",
"valuesPrinted": False,
}
version = status.get("version")
required = status.get("required") is True
if not required:
return {
"ok": True,
"action": "already-acknowledged",
"version": version,
"requiredBefore": False,
"requiredAfter": False,
"phrasePrinted": False,
"valuesPrinted": False,
}
phrase = status.get("ack_phrase_zh") if isinstance(status.get("ack_phrase_zh"), str) and status.get("ack_phrase_zh") else status.get("ack_phrase_en")
language = "zh" if isinstance(status.get("ack_phrase_zh"), str) and status.get("ack_phrase_zh") else "en"
if not isinstance(phrase, str) or not phrase:
raise RuntimeError("admin compliance acknowledgement phrase missing")
accepted = ensure_success(
curl_api("POST", "/api/v1/admin/compliance/accept", bearer=token, payload={"phrase": phrase, "language": language}),
"accept admin compliance",
)
required_after = accepted.get("required") if isinstance(accepted, dict) else None
return {
"ok": required_after is False,
"action": "accepted",
"version": version,
"requiredBefore": True,
"requiredAfter": required_after,
"phrasePrinted": False,
"valuesPrinted": False,
}
def extract_items(data):
if isinstance(data, list):
return data
if isinstance(data, dict):
if isinstance(data.get("items"), list):
return data["items"]
for key in ("groups", "accounts", "api_keys", "keys"):
if isinstance(data.get(key), list):
return data[key]
return []
def find_access_token(data):
if isinstance(data, dict):
for key in ("access_token", "token"):
if isinstance(data.get(key), str) and data[key]:
return data[key]
for value in data.values():
token = find_access_token(value)
if token:
return token
return None
def login():
admin_email = get_config_value("sub2api-config", "ADMIN_EMAIL") or POOL_ADMIN_EMAIL_DEFAULT
admin_password = decode_secret_value(APP_SECRET_NAME, "ADMIN_PASSWORD")
if not admin_password:
raise RuntimeError("ADMIN_PASSWORD missing from sub2api-secrets")
data = ensure_success(curl_api("POST", "/api/v1/auth/login", payload={"email": admin_email, "password": admin_password}), "admin login")
token = find_access_token(data)
if not token:
raise RuntimeError("admin login response did not contain access_token")
compliance = ensure_admin_compliance(token)
if compliance.get("ok") is not True:
raise RuntimeError("admin compliance acknowledgement failed")
return admin_email, token, compliance
def group_payload():
return {
"name": POOL_GROUP_NAME,
"description": POOL_GROUP_DESCRIPTION,
"platform": "openai",
"rate_multiplier": 1,
"is_exclusive": False,
"subscription_type": "standard",
"allow_messages_dispatch": True,
"require_oauth_only": False,
"require_privacy_set": False,
"rpm_limit": 0,
}
def list_groups(token):
data = ensure_success(curl_api("GET", "/api/v1/admin/groups/all?platform=openai", bearer=token), "list groups")
return extract_items(data)
def ensure_group(token):
existing = next((item for item in list_groups(token) if item.get("name") == POOL_GROUP_NAME), None)
payload = group_payload()
if existing is None:
created = ensure_success(curl_api("POST", "/api/v1/admin/groups", bearer=token, payload=payload), "create group")
return created, "created"
group_id = existing.get("id")
if group_id is None:
raise RuntimeError("existing group has no id")
payload["status"] = "active"
updated = ensure_success(curl_api("PUT", f"/api/v1/admin/groups/{group_id}", bearer=token, payload=payload), "update group")
return updated if isinstance(updated, dict) else existing, "updated"
def list_accounts_for_group(token, group_id):
path = f"/api/v1/admin/accounts?group_id={group_id}&page=1&page_size=500&platform=openai"
data = ensure_success(curl_api("GET", path, bearer=token), f"list accounts for group {group_id}")
return extract_items(data)
def account_group_ids(token, account):
if not isinstance(account, dict) or account.get("id") is None:
return []
account_id = account.get("id")
account_name = account.get("name")
ids = []
for group in list_groups(token):
group_id = group.get("id") if isinstance(group, dict) else None
if group_id is None:
continue
members = list_accounts_for_group(token, group_id)
if any(item.get("id") == account_id or item.get("name") == account_name for item in members if isinstance(item, dict)):
ids.append(group_id)
return sorted(set(ids))
def list_accounts(token):
path = "/api/v1/admin/accounts?page=1&page_size=200&platform=openai&type=apikey&search=" + quote("unidesk-codex-")
data = ensure_success(curl_api("GET", path, bearer=token), "list accounts")
return extract_items(data)
def find_account_by_name(token, name):
path = "/api/v1/admin/accounts?page=1&page_size=50&platform=openai&search=" + quote(str(name))
data = ensure_success(curl_api("GET", path, bearer=token), "find account " + str(name))
for item in extract_items(data):
if isinstance(item, dict) and item.get("name") == name:
return item
return None
def list_proxy_candidates(token, search=""):
path = "/api/v1/admin/proxies?page=1&page_size=200"
if search:
path += "&search=" + quote(str(search))
data = ensure_success(curl_api("GET", path, bearer=token), "list proxies")
return extract_items(data)
def find_proxy_by_name(token, name):
for item in list_proxy_candidates(token, name):
if isinstance(item, dict) and item.get("name") == name:
return item
return None
def manual_binding_plan(binding, expected_kind):
if not isinstance(binding, dict):
return None
plan = binding.get("sourcePlan")
if not isinstance(plan, dict):
raise RuntimeError("manual account binding is missing resolved sourcePlan")
if plan.get("enabled") is not True:
raise RuntimeError(f"manual account binding source {plan.get('id')} is disabled")
if plan.get("kind") != expected_kind:
raise RuntimeError(f"manual account binding source {plan.get('id')} kind must be {expected_kind}")
return plan
def desired_manual_proxy_payload(protection):
binding = protection.get("proxyBinding") if isinstance(protection, dict) else None
if not isinstance(binding, dict) or binding.get("enabled") is not True:
return None
plan = manual_binding_plan(binding, "proxy")
if plan.get("provider") != "target-egress-proxy":
raise RuntimeError(f"manual account proxyBinding source {plan.get('id')} uses unsupported provider {plan.get('provider')}")
target_proxy = plan.get("targetEgressProxy")
if not isinstance(target_proxy, dict):
raise RuntimeError(f"manual account proxyBinding source {plan.get('id')} requires an enabled target egressProxy")
service_name = target_proxy.get("serviceName")
listen_port = target_proxy.get("listenPort")
namespace = target_proxy.get("namespace") if isinstance(target_proxy.get("namespace"), str) and target_proxy.get("namespace") else NAMESPACE
proxy_name = binding.get("proxyName")
if not isinstance(service_name, str) or not service_name:
raise RuntimeError("target egressProxy serviceName is missing")
if not isinstance(listen_port, int) or listen_port <= 0:
raise RuntimeError("target egressProxy listenPort is missing")
if not isinstance(proxy_name, str) or not proxy_name:
raise RuntimeError("manual account proxyBinding proxyName is missing")
return {
"name": proxy_name,
"protocol": "http",
"host": f"{service_name}.{namespace}.svc.cluster.local",
"port": listen_port,
"fallback_mode": "none",
"expiry_warn_days": 0,
}
def proxy_needs_update(proxy, payload):
if not isinstance(proxy, dict):
return True
for key in ("name", "protocol", "host", "port", "fallback_mode", "expiry_warn_days"):
if proxy.get(key) != payload.get(key):
return True
return proxy.get("status") != "active"
def ensure_manual_proxy(token, payload):
current = find_proxy_by_name(token, payload["name"])
if current is None:
created = ensure_success(curl_api("POST", "/api/v1/admin/proxies", bearer=token, payload=payload), f"create proxy {payload['name']}")
return created, "created"
if proxy_needs_update(current, payload):
update_payload = dict(payload)
update_payload["status"] = "active"
updated = ensure_success(curl_api("PUT", f"/api/v1/admin/proxies/{current['id']}", bearer=token, payload=update_payload), f"update proxy {payload['name']}")
return updated if isinstance(updated, dict) else current, "updated"
return current, "unchanged"
def manual_proxy_status(token, account, protection):
payload = desired_manual_proxy_payload(protection)
if payload is None:
return {
"enabled": False,
"ok": True,
"action": "not-configured",
"valuesPrinted": False,
}
binding = protection.get("proxyBinding") if isinstance(protection, dict) else None
plan = manual_binding_plan(binding, "proxy")
proxy = find_proxy_by_name(token, payload["name"])
runtime_proxy = account.get("proxy") if isinstance(account, dict) and isinstance(account.get("proxy"), dict) else None
binding_aligned = (
isinstance(account, dict)
and isinstance(proxy, dict)
and account.get("proxy_id") == proxy.get("id")
)
return {
"enabled": True,
"ok": binding_aligned,
"action": "validate",
"source": plan.get("id"),
"sourceProvider": plan.get("provider"),
"expectedProxyName": payload["name"],
"expectedProtocol": payload["protocol"],
"expectedHost": payload["host"],
"expectedPort": payload["port"],
"proxyRecordExists": isinstance(proxy, dict),
"proxyId": proxy.get("id") if isinstance(proxy, dict) else None,
"runtimeProxyId": account.get("proxy_id") if isinstance(account, dict) else None,
"runtimeProxyName": runtime_proxy.get("name") if isinstance(runtime_proxy, dict) else None,
"bindingAligned": binding_aligned,
"valuesPrinted": False,
}
def ensure_manual_account_proxy_bindings(token):
items = []
for protection in MANUAL_ACCOUNT_PROTECTIONS:
if not isinstance(protection, dict):
continue
name = protection.get("accountName")
if not isinstance(name, str) or not name:
continue
account = find_account_by_name(token, name)
payload = desired_manual_proxy_payload(protection)
if payload is None:
items.append({
"accountName": name,
"enabled": False,
"action": "not-configured",
"ok": True,
"valuesPrinted": False,
})
continue
if not isinstance(account, dict):
items.append({
"accountName": name,
"enabled": True,
"action": "account-missing",
"ok": False,
"expectedProxyName": payload["name"],
"valuesPrinted": False,
})
continue
extra = account.get("extra") if isinstance(account.get("extra"), dict) else {}
if extra.get("unidesk_managed") is True:
items.append({
"accountName": name,
"accountId": account.get("id"),
"enabled": True,
"action": "refused-managed-runtime-account",
"ok": False,
"expectedProxyName": payload["name"],
"valuesPrinted": False,
})
continue
proxy, proxy_action = ensure_manual_proxy(token, payload)
binding = protection.get("proxyBinding") if isinstance(protection, dict) else None
plan = manual_binding_plan(binding, "proxy")
proxy_id = proxy.get("id") if isinstance(proxy, dict) else None
if proxy_id is None:
raise RuntimeError(f"proxy {payload['name']} has no id")
action = "unchanged"
if account.get("proxy_id") != proxy_id:
updated = ensure_success(curl_api("PUT", f"/api/v1/admin/accounts/{account['id']}", bearer=token, payload={"proxy_id": proxy_id}), f"bind manual account proxy {name}")
account = updated if isinstance(updated, dict) else account
action = "bound"
runtime_proxy = account.get("proxy") if isinstance(account.get("proxy"), dict) else None
items.append({
"accountName": name,
"accountId": account.get("id"),
"enabled": True,
"action": action,
"proxyAction": proxy_action,
"source": plan.get("id"),
"sourceProvider": plan.get("provider"),
"ok": account.get("proxy_id") == proxy_id,
"expectedProxyName": payload["name"],
"expectedProtocol": payload["protocol"],
"expectedHost": payload["host"],
"expectedPort": payload["port"],
"proxyId": proxy_id,
"runtimeProxyId": account.get("proxy_id"),
"runtimeProxyName": runtime_proxy.get("name") if isinstance(runtime_proxy, dict) else None,
"bindingAligned": account.get("proxy_id") == proxy_id,
"controlPolicy": "manual-protected: only proxy_id binding is YAML-controlled; credentials/status/schedulable are untouched",
"valuesPrinted": False,
})
return {
"ok": all(item.get("ok") is True for item in items),
"itemCount": len(items),
"items": items,
"valuesPrinted": False,
}
def manual_group_binding_plan(protection):
binding = protection.get("groupBinding") if isinstance(protection, dict) else None
if not isinstance(binding, dict) or binding.get("enabled") is not True:
return None
plan = manual_binding_plan(binding, "group")
if plan.get("provider") != "pool-group":
raise RuntimeError(f"manual account groupBinding source {plan.get('id')} uses unsupported provider {plan.get('provider')}")
return plan
def manual_group_binding_enabled(protection):
return manual_group_binding_plan(protection) is not None
def manual_group_status(token, account, protection, group_id):
plan = manual_group_binding_plan(protection)
if plan is None:
return {
"enabled": False,
"ok": True,
"action": "not-configured",
"valuesPrinted": False,
}
group_accounts = list_accounts_for_group(token, group_id)
account_id = account.get("id") if isinstance(account, dict) else None
account_name = account.get("name") if isinstance(account, dict) else None
binding_aligned = any(
item.get("id") == account_id or item.get("name") == account_name
for item in group_accounts
if isinstance(item, dict)
)
return {
"enabled": True,
"ok": binding_aligned,
"action": "validate",
"source": plan.get("id"),
"sourceProvider": plan.get("provider"),
"poolGroupName": POOL_GROUP_NAME,
"poolGroupId": group_id,
"bindingAligned": binding_aligned,
"valuesPrinted": False,
}
def ensure_manual_account_group_bindings(token, group_id):
items = []
for protection in MANUAL_ACCOUNT_PROTECTIONS:
if not isinstance(protection, dict):
continue
name = protection.get("accountName")
if not isinstance(name, str) or not name:
continue
if not manual_group_binding_enabled(protection):
items.append({
"accountName": name,
"enabled": False,
"action": "not-configured",
"ok": True,
"valuesPrinted": False,
})
continue
plan = manual_group_binding_plan(protection)
account = find_account_by_name(token, name)
if not isinstance(account, dict):
items.append({
"accountName": name,
"enabled": True,
"action": "account-missing",
"ok": False,
"poolGroupName": POOL_GROUP_NAME,
"poolGroupId": group_id,
"valuesPrinted": False,
})
continue
extra = account.get("extra") if isinstance(account.get("extra"), dict) else {}
if extra.get("unidesk_managed") is True:
items.append({
"accountName": name,
"accountId": account.get("id"),
"enabled": True,
"action": "refused-managed-runtime-account",
"ok": False,
"poolGroupName": POOL_GROUP_NAME,
"poolGroupId": group_id,
"valuesPrinted": False,
})
continue
existing_group_ids = account_group_ids(token, account)
desired_group_ids = sorted(set(existing_group_ids + [group_id]))
action = "unchanged"
if group_id not in existing_group_ids:
updated = ensure_success(curl_api("PUT", f"/api/v1/admin/accounts/{account['id']}", bearer=token, payload={"group_ids": desired_group_ids}), f"bind manual account group {name}")
account = updated if isinstance(updated, dict) else account
action = "bound"
binding_aligned = any(
item.get("id") == account.get("id") or item.get("name") == name
for item in list_accounts_for_group(token, group_id)
if isinstance(item, dict)
)
items.append({
"accountName": name,
"accountId": account.get("id"),
"enabled": True,
"ok": binding_aligned,
"action": action,
"source": plan.get("id") if isinstance(plan, dict) else None,
"sourceProvider": plan.get("provider") if isinstance(plan, dict) else None,
"poolGroupName": POOL_GROUP_NAME,
"poolGroupId": group_id,
"previousGroupIds": existing_group_ids,
"desiredGroupIds": desired_group_ids,
"bindingAligned": binding_aligned,
"controlPolicy": "manual-protected: only pool group membership is YAML-controlled; credentials/status/schedulable are untouched and sentinel does not probe it",
"valuesPrinted": False,
})
return {
"ok": all(item.get("ok") is True for item in items),
"itemCount": len(items),
"items": items,
"valuesPrinted": False,
}
def manual_account_protection_status(token, group_id=None):
items = []
desired_names = set(EXPECTED_ACCOUNT_CAPACITIES.keys())
for protection in MANUAL_ACCOUNT_PROTECTIONS:
if not isinstance(protection, dict):
continue
name = protection.get("accountName")
if not isinstance(name, str) or not name:
continue
account = find_account_by_name(token, name)
extra = account.get("extra") if isinstance(account, dict) and isinstance(account.get("extra"), dict) else {}
proxy_status = manual_proxy_status(token, account, protection)
group_status = manual_group_status(token, account, protection, group_id) if group_id is not None else {"enabled": False, "ok": True, "action": "not-checked", "valuesPrinted": False}
items.append({
"accountName": name,
"reason": protection.get("reason") if isinstance(protection.get("reason"), str) else None,
"exists": isinstance(account, dict),
"accountId": account.get("id") if isinstance(account, dict) else None,
"status": account.get("status") if isinstance(account, dict) else None,
"schedulable": account.get("schedulable") if isinstance(account, dict) else None,
"inYamlProfiles": name in desired_names,
"runtimeMarkedUnideskManaged": extra.get("unidesk_managed") is True,
"proxyBinding": proxy_status,
"groupBinding": group_status,
"ok": proxy_status.get("ok") is True and group_status.get("ok") is True,
"controlPolicy": "manual-protected: no create/update/prune/probe/freeze; optional proxy_id and pool group membership binding only when configured",
"valuesPrinted": False,
})
return {
"ok": all(item.get("ok") is True for item in items),
"protectedCount": len(items),
"items": items,
"valuesPrinted": False,
}
def list_probe_accounts(token):
path = "/api/v1/admin/accounts?page=1&page_size=200&platform=openai&type=apikey&search=" + quote("unidesk-probe-")
data = ensure_success(curl_api("GET", path, bearer=token), "list probe accounts")
return [item for item in extract_items(data) if isinstance(item.get("name"), str) and item.get("name").startswith("unidesk-probe-")]
def list_probe_groups(token):
data = ensure_success(curl_api("GET", "/api/v1/admin/groups/all?platform=openai", bearer=token), "list groups")
return [item for item in extract_items(data) if isinstance(item.get("name"), str) and item.get("name").startswith("unidesk-probe-")]
def cleanup_probe_resources(token):
api_key_results = []
account_results = []
group_results = []
for item in list_user_keys(token):
name = item.get("name")
key_id = item.get("id")
if not isinstance(name, str) or not name.startswith("unidesk-probe-") or key_id is None:
continue
api_key_results.append(delete_probe_resource(token, "DELETE", f"/api/v1/keys/{key_id}", "api-key"))
for item in list_probe_accounts(token):
account_id = item.get("id")
if account_id is None:
continue
account_results.append(delete_probe_resource(token, "DELETE", f"/api/v1/admin/accounts/{account_id}", "account"))
for item in list_probe_groups(token):
group_id = item.get("id")
if group_id is None:
continue
group_results.append(delete_probe_resource(token, "DELETE", f"/api/v1/admin/groups/{group_id}", "group"))
return {
"ok": all(item.get("ok") is True for item in api_key_results + account_results + group_results),
"apiKeysDeleted": len(api_key_results),
"accountsDeleted": len(account_results),
"groupsDeleted": len(group_results),
"items": {
"apiKeys": api_key_results,
"accounts": account_results,
"groups": group_results,
},
"valuesPrinted": False,
}
def temp_unschedulable_credentials(profile):
credentials = profile.get("tempUnschedulableCredentials")
if not isinstance(credentials, dict):
credentials = {}
return normalize_temp_unschedulable_credentials(credentials)
def account_payload(profile, group_id):
extra = {
"openai_responses_mode": "force_responses",
"unidesk_codex_profile": profile["profile"],
"unidesk_managed": True,
"unidesk_trust_upstream": profile.get("trustUpstream") is True,
"unidesk_sentinel_protect": profile.get("sentinelProtect") if isinstance(profile.get("sentinelProtect"), dict) else {"enabled": False},
}
ws_mode = profile.get("openaiResponsesWebSocketsV2Mode")
if ws_mode:
extra["openai_apikey_responses_websockets_v2_mode"] = ws_mode
extra["openai_apikey_responses_websockets_v2_enabled"] = ws_mode != "off"
credentials = {
"api_key": profile["apiKey"],
"base_url": profile["baseUrl"],
}
upstream_user_agent = profile.get("upstreamUserAgent")
if upstream_user_agent:
credentials["user_agent"] = upstream_user_agent
temp_unschedulable = temp_unschedulable_credentials(profile)
if temp_unschedulable["enabled"]:
credentials["temp_unschedulable_enabled"] = True
credentials["temp_unschedulable_rules"] = temp_unschedulable["rules"]
return {
"name": profile["accountName"],
"notes": f"UniDesk-managed Codex profile {profile['profile']} from {profile['configFile']} and {profile['authFile']}; secret source={profile['apiKeySource']}; fingerprint={profile['apiKeyFingerprint']}.",
"platform": "openai",
"type": "apikey",
"credentials": credentials,
"extra": extra,
"concurrency": int(profile.get("capacity", 5) or 5),
"priority": int(profile.get("priority", POOL_DEFAULT_ACCOUNT_PRIORITY) or POOL_DEFAULT_ACCOUNT_PRIORITY),
"rate_multiplier": 1,
"load_factor": int(profile.get("loadFactor", POOL_DEFAULT_ACCOUNT_LOAD_FACTOR) or POOL_DEFAULT_ACCOUNT_LOAD_FACTOR),
"group_ids": [group_id],
"confirm_mixed_channel_risk": True,
}
def planned_sentinel_account_results(profiles, existing_accounts):
existing = {item.get("name"): item for item in existing_accounts if isinstance(item, dict)}
results = []
for profile in profiles:
change_reasons = sentinel_probe_change_reasons(existing.get(profile["accountName"]), profile)
quality_gate_required = sentinel_quality_gate_enabled() and len(change_reasons) > 0
results.append({
"profile": profile["profile"],
"accountName": profile["accountName"],
"profileConfig": {
"accountName": profile["accountName"],
"trustUpstream": profile.get("trustUpstream") is True,
"sentinelProtect": profile.get("sentinelProtect") if isinstance(profile.get("sentinelProtect"), dict) else {"enabled": False},
},
"sentinelProbeConfigFingerprint": profile.get("sentinelProbeConfigFingerprint"),
"sentinelProbeRequired": quality_gate_required,
"sentinelChangeReasons": change_reasons if quality_gate_required else [],
"sentinelProbePending": quality_gate_required,
"sentinelDefaultFrozen": False,
"valuesPrinted": False,
})
return results
def ensure_accounts(token, profiles, group_id, prune_removed=False, protected_frozen_names=None, existing_accounts=None):
if not isinstance(protected_frozen_names, set):
protected_frozen_names = set()
if not isinstance(existing_accounts, list):
existing_accounts = list_accounts(token)
existing = {item.get("name"): item for item in existing_accounts}
desired_names = {profile["accountName"] for profile in profiles}
results = []
for profile in profiles:
payload = account_payload(profile, group_id)
current = existing.get(profile["accountName"])
change_reasons = sentinel_probe_change_reasons(current, profile)
quality_gate_required = sentinel_quality_gate_enabled() and len(change_reasons) > 0
keep_frozen = profile["accountName"] in protected_frozen_names
if current and current.get("id") is not None:
account_id = current["id"]
update_payload = dict(payload)
update_payload.pop("platform", None)
update_payload["status"] = "active"
data = ensure_success(curl_api("PUT", f"/api/v1/admin/accounts/{account_id}", bearer=token, payload=update_payload), f"update account {profile['accountName']}")
action = "updated"
else:
data = ensure_success(curl_api("POST", "/api/v1/admin/accounts", bearer=token, payload=payload), f"create account {profile['accountName']}")
action = "created"
results.append({
"profile": profile["profile"],
"accountName": profile["accountName"],
"profileConfig": {
"accountName": profile["accountName"],
"trustUpstream": profile.get("trustUpstream") is True,
"sentinelProtect": profile.get("sentinelProtect") if isinstance(profile.get("sentinelProtect"), dict) else {"enabled": False},
},
"accountId": data.get("id") if isinstance(data, dict) else None,
"action": action,
"baseUrl": profile["baseUrl"],
"apiKeySource": profile["apiKeySource"],
"apiKeyFingerprint": profile["apiKeyFingerprint"],
"sentinelProbeConfigFingerprint": profile.get("sentinelProbeConfigFingerprint"),
"sentinelProbeRequired": quality_gate_required,
"sentinelChangeReasons": change_reasons if quality_gate_required else [],
"sentinelProbePending": quality_gate_required,
"sentinelDefaultFrozen": False,
"sentinelFreezeProtected": keep_frozen,
"schedulableControl": "sentinel-marker-restore",
"openaiResponsesWebSocketsV2Mode": profile.get("openaiResponsesWebSocketsV2Mode"),
"trustUpstream": profile.get("trustUpstream") is True,
"priority": int(profile.get("priority", POOL_DEFAULT_ACCOUNT_PRIORITY) or POOL_DEFAULT_ACCOUNT_PRIORITY),
"capacity": int(profile.get("capacity", 5) or 5),
"loadFactor": int(profile.get("loadFactor", POOL_DEFAULT_ACCOUNT_LOAD_FACTOR) or POOL_DEFAULT_ACCOUNT_LOAD_FACTOR),
"runtimeConcurrency": data.get("concurrency") if isinstance(data, dict) else None,
"runtimeLoadFactor": data.get("load_factor") if isinstance(data, dict) else None,
"runtimeSchedulable": data.get("schedulable") if isinstance(data, dict) else None,
"tempUnschedulableConfigured": bool(payload["credentials"].get("temp_unschedulable_enabled")),
"tempUnschedulableRuleCount": len(payload["credentials"].get("temp_unschedulable_rules") or []),
"upstreamUserAgentConfigured": bool(profile.get("upstreamUserAgent")),
"valuesPrinted": False,
})
prune_results = prune_removed_accounts(token, existing_accounts, desired_names) if prune_removed else []
return results, prune_results
def prune_removed_accounts(token, existing_accounts, desired_names):
results = []
for account in existing_accounts:
name = account.get("name")
account_id = account.get("id")
extra = account.get("extra") if isinstance(account.get("extra"), dict) else {}
if not isinstance(name, str) or not name.startswith("unidesk-codex-"):
continue
if extra.get("unidesk_managed") is not True:
continue
if name in desired_names:
continue
if account_id is None:
raise RuntimeError(f"removed account {name} has no id")
ensure_success(curl_api("DELETE", f"/api/v1/admin/accounts/{account_id}", bearer=token), f"delete removed account {name}")
results.append({
"accountName": name,
"accountId": account_id,
"profile": extra.get("unidesk_codex_profile"),
"action": "deleted",
"reason": "removed-from-yaml",
"valuesPrinted": False,
})
return results
def generate_api_key():
alphabet = string.ascii_letters + string.digits
return "sk-unidesk-codex-" + "".join(secrets.choice(alphabet) for _ in range(48))
def ensure_api_key_secret(group_id):
existing = None
try:
existing = decode_secret_value(POOL_API_KEY_SECRET_NAME, POOL_API_KEY_SECRET_KEY)
except Exception:
existing = None
api_key = existing if existing else generate_api_key()
secret_action = "kept-existing" if existing else "created"
manifest = {
"apiVersion": "v1",
"kind": "Secret",
"metadata": {
"name": POOL_API_KEY_SECRET_NAME,
"namespace": NAMESPACE,
"labels": {
"app.kubernetes.io/name": "sub2api",
"app.kubernetes.io/part-of": "platform-infra",
"app.kubernetes.io/managed-by": "unidesk",
"unidesk.ai/secret-purpose": "sub2api-codex-pool-api-key",
},
},
"type": "Opaque",
"stringData": {
POOL_API_KEY_SECRET_KEY: api_key,
"GROUP_ID": str(group_id),
"GROUP_NAME": POOL_GROUP_NAME,
"SERVICE_DNS": SERVICE_DNS,
},
}
proc = kubectl(["apply", "--server-side", "--force-conflicts", f"--field-manager={FIELD_MANAGER}", "-f", "-"], json.dumps(manifest))
if proc.returncode != 0:
raise RuntimeError(f"apply API key secret failed: {text(proc.stderr, 1000)}")
return api_key, secret_action, text(proc.stdout, 1000)
def apply_sentinel_manifest(manifest):
if not isinstance(manifest, str) or not manifest.strip():
return {
"ok": False,
"action": "missing-manifest",
"valuesPrinted": False,
}
proc = kubectl(["apply", "--server-side", "--force-conflicts", f"--field-manager={FIELD_MANAGER}", "-f", "-"], manifest)
if proc.returncode != 0:
return {
"ok": False,
"action": "apply-failed",
"stdoutTail": text(proc.stdout, 2000),
"stderrTail": text(proc.stderr, 4000),
"valuesPrinted": False,
}
status = sentinel_runtime_status()
return {
"ok": status.get("ok") is True,
"action": "applied",
"stdoutTail": text(proc.stdout, 2000),
"runtime": status,
"valuesPrinted": False,
}
def safe_kube_json(args, label):
proc = kubectl([*args, "-o", "json"])
if proc.returncode != 0:
return None, {"label": label, "exitCode": proc.returncode, "stderrTail": text(proc.stderr, 1000)}
try:
return json.loads(proc.stdout.decode("utf-8")), None
except Exception as exc:
return None, {"label": label, "exitCode": proc.returncode, "error": str(exc), "stdoutTail": text(proc.stdout, 1000)}
def sentinel_runtime_status():
cfg = SENTINEL_CONFIG
cronjob_name = cfg.get("cronJobName")
secret_name = cfg.get("credentialsSecretName")
configmap_name = cfg.get("configMapName")
state_name = cfg.get("stateConfigMapName")
cronjob, cronjob_error = safe_kube_json(["-n", NAMESPACE, "get", "cronjob", cronjob_name], f"cronjob/{cronjob_name}")
secret, secret_error = safe_kube_json(["-n", NAMESPACE, "get", "secret", secret_name], f"secret/{secret_name}")
configmap, configmap_error = safe_kube_json(["-n", NAMESPACE, "get", "configmap", configmap_name], f"configmap/{configmap_name}")
state_cm, state_error = safe_kube_json(["-n", NAMESPACE, "get", "configmap", state_name], f"configmap/{state_name}")
state = None
if isinstance(state_cm, dict):
raw_state = (state_cm.get("data") or {}).get("state.json")
if isinstance(raw_state, str) and raw_state:
try:
state = json.loads(raw_state)
except Exception as exc:
state = {"parseError": str(exc)}
accounts = (state.get("accounts") or {}) if isinstance(state, dict) else {}
quarantined = []
recent_accounts = []
for name, account_state in accounts.items():
if not isinstance(account_state, dict):
continue
quarantine = account_state.get("quarantine")
if isinstance(quarantine, dict) and quarantine.get("active") is True:
quarantined.append({
"accountName": name,
"until": quarantine.get("until"),
"applied": quarantine.get("applied"),
"reason": quarantine.get("reason"),
"failureKind": quarantine.get("failureKind"),
"errorDetails": quarantine.get("errorDetails"),
"intervalMinutes": quarantine.get("intervalMinutes"),
"sentinelProtect": quarantine.get("sentinelProtect"),
})
last_probe = account_state.get("lastProbe")
if isinstance(last_probe, dict):
recent_accounts.append({
"accountName": name,
"lastProbeAt": account_state.get("lastProbeAt"),
"lastStatus": account_state.get("lastStatus"),
"nextProbeAfter": account_state.get("nextProbeAfter"),
"ok": last_probe.get("ok"),
"purpose": last_probe.get("purpose"),
"httpStatus": last_probe.get("httpStatus"),
"durationMs": last_probe.get("durationMs"),
"markerMatched": last_probe.get("markerMatched"),
"sentinelProtect": last_probe.get("sentinelProtect") or account_state.get("sentinelProtectConfig"),
"outputHash": last_probe.get("outputHash"),
"outputPreview": last_probe.get("outputPreview"),
"responseBodyHash": last_probe.get("responseBodyHash"),
"responseBodyPreview": last_probe.get("responseBodyPreview"),
"error": last_probe.get("error"),
"errorDetails": last_probe.get("errorDetails"),
"usage": last_probe.get("usage"),
"failureKind": last_probe.get("failureKind"),
"requestShape": last_probe.get("requestShape"),
"action": last_probe.get("action"),
})
recent_accounts.sort(key=lambda item: item.get("lastProbeAt") or "")
last_run = state.get("lastRun") if isinstance(state, dict) else None
cronjob_spec = cronjob.get("spec") if isinstance(cronjob, dict) else {}
secret_data = secret.get("data") if isinstance(secret, dict) else {}
configmap_data = configmap.get("data") if isinstance(configmap, dict) else {}
ok = cronjob is not None and secret is not None and configmap is not None
return {
"ok": ok,
"desired": {
"monitorEnabled": cfg.get("monitor", {}).get("enabled"),
"actionsEnabled": cfg.get("actions", {}).get("enabled"),
"schedule": cfg.get("schedule"),
"cronJobName": cronjob_name,
"configMapName": configmap_name,
"credentialsSecretName": secret_name,
"stateConfigMapName": state_name,
},
"cronJob": {
"exists": cronjob is not None,
"schedule": cronjob_spec.get("schedule") if isinstance(cronjob_spec, dict) else None,
"suspend": cronjob_spec.get("suspend") if isinstance(cronjob_spec, dict) else None,
"lastScheduleTime": (cronjob.get("status") or {}).get("lastScheduleTime") if isinstance(cronjob, dict) else None,
"active": len((cronjob.get("status") or {}).get("active") or []) if isinstance(cronjob, dict) else None,
"error": cronjob_error,
},
"secret": {
"exists": secret is not None,
"profileSecretPresent": isinstance(secret_data, dict) and "profiles.json" in secret_data,
"valuesPrinted": False,
"error": secret_error,
},
"configMap": {
"exists": configmap is not None,
"configPresent": isinstance(configmap_data, dict) and "config.json" in configmap_data,
"runnerPresent": isinstance(configmap_data, dict) and "sentinel.py" in configmap_data,
"error": configmap_error,
},
"state": {
"exists": state_cm is not None,
"accountCount": len(accounts),
"quarantinedCount": len(quarantined),
"quarantined": quarantined[-10:],
"recentAccounts": recent_accounts[-12:],
"lastRun": last_run,
"error": state_error,
},
"valuesPrinted": False,
}
def parse_epoch_z(value):
if not isinstance(value, str) or not value:
return None
try:
return datetime.fromisoformat(value.replace("Z", "+00:00")).timestamp()
except Exception:
return None
def sentinel_state_object():
state_name = SENTINEL_CONFIG.get("stateConfigMapName")
if not state_name:
return None, None
obj, err = safe_kube_json(["-n", NAMESPACE, "get", "configmap", state_name], f"configmap/{state_name}")
if not isinstance(obj, dict):
return None, None
raw_state = (obj.get("data") or {}).get("state.json")
if not isinstance(raw_state, str) or not raw_state:
return obj, None
try:
return obj, json.loads(raw_state)
except Exception:
return obj, None
def active_sentinel_quarantine_names():
_, state = sentinel_state_object()
if not isinstance(state, dict):
return set()
accounts_state = state.get("accounts") if isinstance(state.get("accounts"), dict) else {}
names = set()
for name, account_state in accounts_state.items():
if not isinstance(name, str) or not isinstance(account_state, dict):
continue
quarantine = account_state.get("quarantine")
if isinstance(quarantine, dict) and quarantine.get("active") is True and quarantine.get("applied") is True:
names.add(name)
return names
def default_sentinel_state():
return {"version": 1, "accounts": {}, "ledger": {}, "history": []}
def clamp_sentinel_freezes_for_config(state, now):
freeze_config = SENTINEL_CONFIG.get("freeze") if isinstance(SENTINEL_CONFIG.get("freeze"), dict) else {}
try:
max_interval = int(freeze_config.get("maxTtlMinutes") or 10)
except Exception:
max_interval = 10
accounts_state = state.get("accounts") if isinstance(state.get("accounts"), dict) else {}
now_epoch = time.time()
items = []
for name, account_state in accounts_state.items():
if not isinstance(name, str) or not isinstance(account_state, dict):
continue
quarantine = account_state.get("quarantine")
if not isinstance(quarantine, dict) or quarantine.get("active") is not True or quarantine.get("applied") is not True:
continue
try:
interval = int(quarantine.get("intervalMinutes") or 0)
except Exception:
interval = 0
until_epoch = parse_epoch_z(quarantine.get("until"))
old_until = quarantine.get("until")
if interval <= max_interval and (until_epoch is None or until_epoch <= now_epoch + max_interval * 60):
continue
quarantine["previousIntervalMinutes"] = interval
quarantine["intervalMinutes"] = max_interval
quarantine["until"] = now
quarantine["clampedAt"] = now
quarantine["clampedBy"] = "sync-freeze-max-ttl"
account_state["nextProbeAfter"] = now
items.append({
"accountName": name,
"previousIntervalMinutes": interval,
"maxIntervalMinutes": max_interval,
"previousUntil": old_until,
"nextProbeAfter": now,
})
return items
def parse_iso_epoch(value):
if not isinstance(value, str) or not value:
return None
try:
return datetime.fromisoformat(value.replace("Z", "+00:00")).timestamp()
except Exception:
return None
def profile_success_max_interval(profile):
cadence = SENTINEL_CONFIG.get("cadence") if isinstance(SENTINEL_CONFIG.get("cadence"), dict) else {}
legacy = cadence.get("successMaxIntervalMinutes")
if legacy is None:
legacy = cadence.get("trustedSuccessMaxIntervalMinutes") or cadence.get("untrustedSuccessMaxIntervalMinutes") or 1
if profile.get("trustUpstream") is True:
value = cadence.get("trustedSuccessMaxIntervalMinutes") or legacy
else:
value = cadence.get("untrustedSuccessMaxIntervalMinutes") or legacy
try:
return int(value)
except Exception:
return int(legacy)
def clamp_sentinel_success_cadence_for_config(state, profiles, now):
accounts_state = state.get("accounts") if isinstance(state.get("accounts"), dict) else {}
profile_map = {item.get("accountName"): item for item in profiles if isinstance(item, dict) and isinstance(item.get("accountName"), str)}
now_epoch = time.time()
items = []
for name, profile in profile_map.items():
account_state = accounts_state.get(name)
if not isinstance(account_state, dict):
continue
quarantine = account_state.get("quarantine")
if isinstance(quarantine, dict) and quarantine.get("active") is True:
account_state["trustUpstream"] = profile.get("trustUpstream") is True
account_state["sentinelProtectConfig"] = profile.get("sentinelProtect") if isinstance(profile.get("sentinelProtect"), dict) else {"enabled": False}
account_state["successMaxIntervalMinutes"] = profile_success_max_interval(profile)
continue
try:
interval = int(account_state.get("successIntervalMinutes") or 0)
except Exception:
interval = 0
next_epoch = parse_iso_epoch(account_state.get("nextProbeAfter"))
max_interval = profile_success_max_interval(profile)
account_state["trustUpstream"] = profile.get("trustUpstream") is True
account_state["sentinelProtectConfig"] = profile.get("sentinelProtect") if isinstance(profile.get("sentinelProtect"), dict) else {"enabled": False}
account_state["successMaxIntervalMinutes"] = max_interval
if interval <= max_interval and (next_epoch is None or next_epoch <= now_epoch + max_interval * 60):
continue
old_next = account_state.get("nextProbeAfter")
account_state["previousSuccessIntervalMinutes"] = interval
account_state["successIntervalMinutes"] = min(interval, max_interval) if interval > 0 else interval
account_state["nextProbeAfter"] = now
account_state["cadenceClampedAt"] = now
account_state["cadenceClampedBy"] = "sync-success-max-interval"
items.append({
"accountName": name,
"trustUpstream": profile.get("trustUpstream") is True,
"previousSuccessIntervalMinutes": interval,
"maxIntervalMinutes": max_interval,
"previousNextProbeAfter": old_next,
"nextProbeAfter": now,
})
return items
def update_sentinel_state_configmap(obj, state):
state_name = SENTINEL_CONFIG.get("stateConfigMapName")
if not state_name:
return {"ok": False, "reason": "state-configmap-missing"}
state_json = json.dumps(state, ensure_ascii=False, indent=2)
manifest = {
"apiVersion": "v1",
"kind": "ConfigMap",
"metadata": {
"name": state_name,
"namespace": NAMESPACE,
"labels": {
"app.kubernetes.io/name": SERVICE_NAME,
"app.kubernetes.io/component": "account-sentinel",
"app.kubernetes.io/managed-by": "unidesk-platform-infra",
},
},
"data": {"state.json": state_json},
}
proc = kubectl(["apply", "--server-side", "--force-conflicts", f"--field-manager={FIELD_MANAGER}", "-f", "-"], json.dumps(manifest))
action = "applied" if isinstance(obj, dict) else "created"
if proc.returncode != 0:
return {"ok": False, "reason": f"{action}-failed", "stderrTail": text(proc.stderr, 2000)}
return {"ok": True, "action": action}
def ensure_sentinel_state_for_sync(account_results, pending_only=False):
if not sentinel_quality_gate_enabled():
return {"ok": True, "skipped": True, "reason": "sentinel-quality-gate-disabled", "changedCount": 0, "items": [], "valuesPrinted": False}
state_obj, state = sentinel_state_object()
if not isinstance(state, dict):
state = default_sentinel_state()
state.setdefault("version", 1)
accounts_state = state.setdefault("accounts", {})
if not isinstance(accounts_state, dict):
accounts_state = {}
state["accounts"] = accounts_state
now = utc_iso()
items = []
clamped_items = [] if pending_only else clamp_sentinel_freezes_for_config(state, now)
cadence_clamped_items = [] if pending_only else clamp_sentinel_success_cadence_for_config(state, [item.get("profileConfig") for item in account_results if isinstance(item.get("profileConfig"), dict)], now)
changed_count = 0
fingerprint_only_count = 0
for item in account_results:
name = item.get("accountName")
if not isinstance(name, str) or not name:
continue
account_state = accounts_state.setdefault(name, {})
if not isinstance(account_state, dict):
account_state = {}
accounts_state[name] = account_state
fingerprint_value = item.get("sentinelProbeConfigFingerprint")
if isinstance(fingerprint_value, str) and fingerprint_value:
account_state["probeConfigFingerprint"] = fingerprint_value
if item.get("sentinelProbeRequired") is not True:
fingerprint_only_count += 1
continue
changed_count += 1
reasons = item.get("sentinelChangeReasons") if isinstance(item.get("sentinelChangeReasons"), list) else []
quarantine = account_state.get("quarantine") if isinstance(account_state.get("quarantine"), dict) else None
if not (isinstance(quarantine, dict) and quarantine.get("active") is True):
account_state["quarantine"] = {
"active": False,
"reason": "yaml-account-change-pending-sentinel-probe",
"lastPendingAt": now,
"changeReasons": reasons,
}
account_state["nextProbeAfter"] = now
account_state["successStreak"] = 0
account_state["successIntervalMinutes"] = 0
profile_config = item.get("profileConfig") if isinstance(item.get("profileConfig"), dict) else {}
account_state["trustUpstream"] = profile_config.get("trustUpstream") is True
account_state["sentinelProtectConfig"] = profile_config.get("sentinelProtect") if isinstance(profile_config.get("sentinelProtect"), dict) else {"enabled": False}
account_state["successMaxIntervalMinutes"] = profile_success_max_interval(profile_config)
account_state["lastStatus"] = "pending-sentinel-quality-gate"
account_state["qualityGate"] = {
"pending": True,
"reason": "yaml-account-change",
"changeReasons": reasons,
"markedAt": now,
"pendingOnly": pending_only,
"defaultFrozen": False,
}
items.append({"accountName": name, "changeReasons": reasons, "nextProbeAfter": now, "defaultFrozen": False, "defaultSchedulable": True, "pendingOnly": pending_only})
if changed_count <= 0 and len(clamped_items) <= 0 and len(cadence_clamped_items) <= 0:
return {"ok": True, "skipped": False, "reason": "no-new-or-changed-accounts", "changedCount": 0, "fingerprintOnlyCount": fingerprint_only_count, "clampedCount": 0, "cadenceClampedCount": 0, "items": [], "valuesPrinted": False}
update = update_sentinel_state_configmap(state_obj, state)
if pending_only and changed_count > 0:
reason = "new-or-changed-accounts-pending-probe-prepared-default-schedulable"
elif changed_count > 0 and (len(clamped_items) > 0 or len(cadence_clamped_items) > 0):
reason = "new-or-changed-accounts-default-schedulable-and-sentinel-cadence-clamped"
elif changed_count > 0:
reason = "new-or-changed-accounts-default-schedulable"
elif len(cadence_clamped_items) > 0:
reason = "success-cadence-clamped-to-current-config"
else:
reason = "freeze-backoff-clamped-to-current-config"
return {
"ok": update.get("ok") is True,
"skipped": False,
"reason": reason,
"changedCount": changed_count,
"fingerprintOnlyCount": fingerprint_only_count,
"clampedCount": len(clamped_items),
"cadenceClampedCount": len(cadence_clamped_items),
"pendingOnly": pending_only,
"items": items,
"clampedItems": clamped_items,
"cadenceClampedItems": cadence_clamped_items,
"update": update,
"valuesPrinted": False,
}
def sentinel_state_summary():
_, state = sentinel_state_object()
if not isinstance(state, dict):
return {"exists": False, "valuesPrinted": False}
accounts = state.get("accounts") if isinstance(state.get("accounts"), dict) else {}
quarantined = []
recent_accounts = []
for name, account_state in accounts.items():
if not isinstance(account_state, dict):
continue
quarantine = account_state.get("quarantine")
if isinstance(quarantine, dict) and quarantine.get("active") is True:
quarantined.append({
"accountName": name,
"until": quarantine.get("until"),
"applied": quarantine.get("applied"),
"reason": quarantine.get("reason"),
"failureKind": quarantine.get("failureKind"),
"errorDetails": quarantine.get("errorDetails"),
"intervalMinutes": quarantine.get("intervalMinutes"),
"sentinelProtect": quarantine.get("sentinelProtect"),
})
last_probe = account_state.get("lastProbe")
if isinstance(last_probe, dict):
recent_accounts.append({
"accountName": name,
"lastProbeAt": account_state.get("lastProbeAt"),
"lastStatus": account_state.get("lastStatus"),
"nextProbeAfter": account_state.get("nextProbeAfter"),
"ok": last_probe.get("ok"),
"purpose": last_probe.get("purpose"),
"httpStatus": last_probe.get("httpStatus"),
"durationMs": last_probe.get("durationMs"),
"markerMatched": last_probe.get("markerMatched"),
"sentinelProtect": last_probe.get("sentinelProtect") or account_state.get("sentinelProtectConfig"),
"usage": last_probe.get("usage"),
"responseBodyHash": last_probe.get("responseBodyHash"),
"responseBodyPreview": last_probe.get("responseBodyPreview"),
"error": last_probe.get("error"),
"errorDetails": last_probe.get("errorDetails"),
"failureKind": last_probe.get("failureKind"),
"requestShape": last_probe.get("requestShape"),
"action": last_probe.get("action"),
})
recent_accounts.sort(key=lambda item: item.get("lastProbeAt") or "")
return {
"exists": True,
"accountCount": len(accounts),
"quarantinedCount": len(quarantined),
"quarantined": quarantined[-10:],
"recentAccounts": recent_accounts[-12:],
"lastRun": state.get("lastRun"),
"valuesPrinted": False,
}
def reassert_sentinel_freezes_after_sync(token):
if (SENTINEL_CONFIG.get("actions") or {}).get("enabled") is not True:
return {"ok": True, "skipped": True, "reason": "actions-disabled", "items": [], "valuesPrinted": False}
_, state = sentinel_state_object()
if not isinstance(state, dict):
return {"ok": True, "skipped": True, "reason": "state-missing", "items": [], "valuesPrinted": False}
accounts_state = state.get("accounts") if isinstance(state.get("accounts"), dict) else {}
accounts = list_accounts(token)
by_name = {item.get("name"): item for item in accounts if isinstance(item.get("name"), str)}
items = []
for name, account_state in accounts_state.items():
if not isinstance(account_state, dict):
continue
quarantine = account_state.get("quarantine")
if not isinstance(quarantine, dict) or quarantine.get("active") is not True or quarantine.get("applied") is not True:
continue
account = by_name.get(name)
if not account or account.get("id") is None:
items.append({"accountName": name, "ok": False, "reason": "account-not-found"})
continue
try:
ensure_success(
curl_api("POST", f"/api/v1/admin/accounts/{account['id']}/schedulable", bearer=token, payload={"schedulable": False}),
f"reassert sentinel freeze for {name}",
)
items.append({"accountName": name, "accountId": account.get("id"), "ok": True, "until": quarantine.get("until")})
except Exception as exc:
items.append({"accountName": name, "accountId": account.get("id"), "ok": False, "error": str(exc)})
return {
"ok": all(item.get("ok") is True for item in items),
"skipped": False,
"items": items,
"valuesPrinted": False,
}
def list_user_keys(token):
data = ensure_success(curl_api("GET", "/api/v1/keys?page=1&page_size=200", bearer=token), "list user keys")
return extract_items(data)
def ensure_sub2api_api_key(token, api_key, group_id):
keys = list_user_keys(token)
existing = next((item for item in keys if item.get("key") == api_key), None)
action = "kept-existing"
if existing is None:
existing = next((item for item in keys if item.get("name") == POOL_API_KEY_NAME), None)
if existing is None or existing.get("key") != api_key:
payload = {
"name": POOL_API_KEY_NAME,
"group_id": group_id,
"custom_key": api_key,
"quota": 0,
"rate_limit_5h": 0,
"rate_limit_1d": 0,
"rate_limit_7d": 0,
}
created = ensure_success(curl_api("POST", "/api/v1/keys", bearer=token, payload=payload), "create pool API key")
existing = created if isinstance(created, dict) else existing
action = "created"
elif existing.get("id") is not None and existing.get("group_id") != group_id:
updated = ensure_success(curl_api("PUT", f"/api/v1/keys/{existing['id']}", bearer=token, payload={"name": POOL_API_KEY_NAME, "group_id": group_id}), "update pool API key group")
existing = updated if isinstance(updated, dict) else existing
action = "updated-group"
return {
"action": action,
"id": existing.get("id") if isinstance(existing, dict) else None,
"name": existing.get("name") if isinstance(existing, dict) else POOL_API_KEY_NAME,
"groupId": existing.get("group_id") if isinstance(existing, dict) else group_id,
"userId": existing.get("user_id") if isinstance(existing, dict) else None,
}
def get_admin_user(token, user_id):
data = ensure_success(curl_api("GET", f"/api/v1/admin/users/{user_id}", bearer=token), "get API key owner")
if not isinstance(data, dict):
raise RuntimeError("API key owner response is not an object")
return data
def ensure_pool_owner_balance(token, user_id):
user = get_admin_user(token, user_id)
current = float(user.get("balance") or 0)
if current >= MIN_OWNER_BALANCE_USD:
return {
"action": "kept-existing",
"userId": user_id,
"balanceBefore": current,
"balanceAfter": current,
"minimumBalanceUsd": MIN_OWNER_BALANCE_USD,
}
updated = ensure_success(curl_api("POST", f"/api/v1/admin/users/{user_id}/balance", bearer=token, payload={
"balance": MIN_OWNER_BALANCE_USD,
"operation": "set",
"notes": "UniDesk Sub2API Codex pool internal API key bootstrap balance.",
}), "set API key owner balance")
after = float(updated.get("balance") or MIN_OWNER_BALANCE_USD) if isinstance(updated, dict) else MIN_OWNER_BALANCE_USD
return {
"action": "set",
"userId": user_id,
"balanceBefore": current,
"balanceAfter": after,
"minimumBalanceUsd": MIN_OWNER_BALANCE_USD,
}
def ensure_pool_owner_concurrency(token, user_id):
user = get_admin_user(token, user_id)
try:
current = int(user.get("concurrency") or 0)
except Exception:
current = 0
if current >= MIN_OWNER_CONCURRENCY:
return {
"ok": True,
"action": "kept-existing",
"userId": user_id,
"concurrencyBefore": current,
"concurrencyAfter": current,
"minimumConcurrency": MIN_OWNER_CONCURRENCY,
"minimumConcurrencySource": MIN_OWNER_CONCURRENCY_SOURCE,
}
updated = ensure_success(curl_api("PUT", f"/api/v1/admin/users/{user_id}", bearer=token, payload={
"concurrency": MIN_OWNER_CONCURRENCY,
}), "set API key owner concurrency")
try:
after = int(updated.get("concurrency") or MIN_OWNER_CONCURRENCY) if isinstance(updated, dict) else MIN_OWNER_CONCURRENCY
except Exception:
after = MIN_OWNER_CONCURRENCY
return {
"ok": after >= MIN_OWNER_CONCURRENCY,
"action": "set",
"userId": user_id,
"concurrencyBefore": current,
"concurrencyAfter": after,
"minimumConcurrency": MIN_OWNER_CONCURRENCY,
"minimumConcurrencySource": MIN_OWNER_CONCURRENCY_SOURCE,
}
def validate_gateway(api_key):
resp = curl_api("GET", "/v1/models", bearer=api_key)
parsed = resp.get("json")
model_count = None
if isinstance(parsed, dict) and isinstance(parsed.get("data"), list):
model_count = len(parsed["data"])
return {
"ok": resp.get("ok"),
"httpStatus": resp.get("httpStatus"),
"transportExitCode": resp.get("transportExitCode"),
"modelCount": model_count,
"bodyPreview": text(resp.get("body", ""), 500) if not resp.get("ok") else "",
"stderr": resp.get("stderr", ""),
"method": "GET /v1/models",
"serviceDns": SERVICE_DNS,
"valuesPrinted": False,
}
def response_output_preview(parsed):
if not isinstance(parsed, dict):
return ""
if isinstance(parsed.get("output_text"), str):
return parsed["output_text"][:240]
output = parsed.get("output")
if not isinstance(output, list):
return ""
parts = []
for item in output:
if not isinstance(item, dict):
continue
content = item.get("content")
if not isinstance(content, list):
continue
for block in content:
if not isinstance(block, dict):
continue
text_value = block.get("text")
if isinstance(text_value, str) and text_value:
parts.append(text_value)
return "\\n".join(parts)[:240]
def request_log_evidence(request_id):
proc = kubectl(["-n", NAMESPACE, "logs", "deployment/sub2api", "--since=5m", "--tail=800"])
stdout = proc.stdout.decode("utf-8", errors="replace")
lines = [line for line in stdout.splitlines() if request_id in line]
failovers = []
final = None
for line in lines:
json_start = line.find("{")
if json_start < 0:
continue
try:
item = json.loads(line[json_start:])
except Exception:
continue
if "upstream_failover_switching" in line:
failovers.append({
"accountId": item.get("account_id"),
"upstreamStatus": item.get("upstream_status"),
"switchCount": item.get("switch_count"),
"maxSwitches": item.get("max_switches"),
})
if "http request completed" in line:
final = {
"accountId": item.get("account_id"),
"statusCode": item.get("status_code"),
"latencyMs": item.get("latency_ms"),
"path": item.get("path"),
}
return {
"requestId": request_id,
"matchedLogLineCount": len(lines),
"failovers": failovers,
"final": final,
"logsExitCode": proc.returncode,
"logsStderr": text(proc.stderr, 1000),
}
def recent_compact_gateway_evidence():
proc = kubectl(["-n", NAMESPACE, "logs", "deployment/sub2api", "--since=6h", "--tail=2500"])
stdout = proc.stdout.decode("utf-8", errors="replace")
failures = []
successes = []
failovers = []
final_errors = []
context_canceled = []
for line in stdout.splitlines():
if "/responses/compact" not in line and "remote_compact" not in line:
continue
json_start = line.find("{")
if json_start < 0:
continue
try:
item = json.loads(line[json_start:])
except Exception:
continue
path = item.get("path")
entry = {
"requestId": item.get("request_id"),
"clientRequestId": item.get("client_request_id"),
"accountId": item.get("account_id"),
"statusCode": item.get("status_code"),
"upstreamStatus": item.get("upstream_status"),
"latencyMs": item.get("latency_ms"),
"path": path,
}
if "codex.remote_compact.failed" in line:
failures.append(entry)
elif "codex.remote_compact.succeeded" in line:
successes.append(entry)
elif "upstream_failover_switching" in line and path == "/responses/compact":
failovers.append({
**entry,
"switchCount": item.get("switch_count"),
"maxSwitches": item.get("max_switches"),
})
elif "http request completed" in line and path == "/responses/compact" and isinstance(item.get("status_code"), int) and item.get("status_code") >= 400:
final_errors.append(entry)
if "context canceled" in line and path == "/responses/compact":
context_canceled.append(entry)
return {
"ok": True,
"degraded": len(failures) > 0 or len(final_errors) > 0 or len(context_canceled) > 0,
"window": "6h",
"tailLines": 2500,
"failureCount": len(failures),
"successCount": len(successes),
"failoverCount": len(failovers),
"finalErrorCount": len(final_errors),
"contextCanceledCount": len(context_canceled),
"recentFailures": failures[-5:],
"recentSuccesses": successes[-5:],
"recentFailovers": failovers[-8:],
"recentFinalErrors": final_errors[-5:],
"recentContextCanceled": context_canceled[-5:],
"logsExitCode": proc.returncode,
"logsStderr": text(proc.stderr, 1000),
"valuesPrinted": False,
}
def is_ignored_probe_noise(entry):
for value in (entry.get("requestId"), entry.get("clientRequestId")):
if isinstance(value, str) and value.startswith("unidesk-400-model-probe-"):
return True
return False
def filter_ignored_probe_noise(items):
return [item for item in items if not is_ignored_probe_noise(item)]
def ignored_probe_noise(items, section):
result = []
for item in items:
if is_ignored_probe_noise(item):
probe = dict(item)
probe["section"] = section
result.append(probe)
return result
def group_by_request_id(items):
grouped = {}
for item in items:
request_id = item.get("requestId")
if not isinstance(request_id, str) or not request_id:
continue
grouped.setdefault(request_id, []).append(item)
return grouped
def failover_budget_exhausted_evidence(failovers, final_errors):
final_by_request = {}
for item in final_errors:
request_id = item.get("requestId")
if isinstance(request_id, str) and request_id:
final_by_request[request_id] = item
exhausted = []
for request_id, request_failovers in group_by_request_id(failovers).items():
final = final_by_request.get(request_id)
if not final:
continue
last = request_failovers[-1]
switch_count = last.get("switchCount")
max_switches = last.get("maxSwitches")
final_status = final.get("statusCode")
if (
isinstance(switch_count, int)
and isinstance(max_switches, int)
and max_switches > 0
and switch_count >= max_switches
and isinstance(final_status, int)
and final_status >= 500
):
exhausted.append({
"requestId": request_id,
"clientRequestId": final.get("clientRequestId") or last.get("clientRequestId"),
"path": final.get("path") or last.get("path"),
"finalAccountId": final.get("accountId"),
"finalStatusCode": final_status,
"switchCount": switch_count,
"maxSwitches": max_switches,
"lastFailoverAccountId": last.get("accountId"),
"lastUpstreamStatus": last.get("upstreamStatus"),
})
return exhausted
def recent_responses_gateway_evidence():
proc = kubectl(["-n", NAMESPACE, "logs", "deployment/sub2api", "--since=6h", "--tail=2500"])
stdout = proc.stdout.decode("utf-8", errors="replace")
failovers = []
forward_failures = []
final_errors = []
context_canceled = []
slow_final_errors = []
for line in stdout.splitlines():
if '"/responses"' not in line and '"/v1/responses"' not in line:
continue
json_start = line.find("{")
if json_start < 0:
continue
try:
item = json.loads(line[json_start:])
except Exception:
continue
path = item.get("path")
if path not in ("/responses", "/v1/responses"):
continue
entry = {
"requestId": item.get("request_id"),
"clientRequestId": item.get("client_request_id"),
"accountId": item.get("account_id"),
"statusCode": item.get("status_code"),
"upstreamStatus": item.get("upstream_status"),
"latencyMs": item.get("latency_ms"),
"path": path,
}
if "upstream_failover_switching" in line:
failovers.append({
**entry,
"switchCount": item.get("switch_count"),
"maxSwitches": item.get("max_switches"),
})
elif "openai.forward_failed" in line:
forward_failures.append({
**entry,
"errorPreview": text(str(item.get("error") or ""), 500),
"fallbackErrorResponseWritten": item.get("fallback_error_response_written"),
"upstreamErrorResponseAlreadyWritten": item.get("upstream_error_response_already_written"),
})
elif "http request completed" in line and isinstance(item.get("status_code"), int) and item.get("status_code") >= 400:
final_errors.append(entry)
latency_ms = item.get("latency_ms")
if isinstance(latency_ms, int) and latency_ms >= 30000:
slow_final_errors.append(entry)
if "context canceled" in line:
context_canceled.append(entry)
visible_failovers = filter_ignored_probe_noise(failovers)
visible_forward_failures = filter_ignored_probe_noise(forward_failures)
visible_final_errors = filter_ignored_probe_noise(final_errors)
visible_slow_final_errors = filter_ignored_probe_noise(slow_final_errors)
visible_context_canceled = filter_ignored_probe_noise(context_canceled)
failover_budget_exhausted = failover_budget_exhausted_evidence(visible_failovers, visible_final_errors)
probe_noise = (
ignored_probe_noise(failovers, "failovers")
+ ignored_probe_noise(forward_failures, "forwardFailures")
+ ignored_probe_noise(final_errors, "finalErrors")
+ ignored_probe_noise(slow_final_errors, "slowFinalErrors")
+ ignored_probe_noise(context_canceled, "contextCanceled")
)
return {
"ok": True,
"degraded": len(visible_forward_failures) > 0 or len(visible_final_errors) > 0 or len(visible_context_canceled) > 0,
"window": "6h",
"tailLines": 2500,
"failoverCount": len(visible_failovers),
"forwardFailureCount": len(visible_forward_failures),
"finalErrorCount": len(visible_final_errors),
"slowFinalErrorCount": len(visible_slow_final_errors),
"contextCanceledCount": len(visible_context_canceled),
"ignoredProbeNoiseCount": len(probe_noise),
"failoverBudgetExhausted": failover_budget_exhausted[-8:],
"rawCounts": {
"failoverCount": len(failovers),
"forwardFailureCount": len(forward_failures),
"finalErrorCount": len(final_errors),
"slowFinalErrorCount": len(slow_final_errors),
"contextCanceledCount": len(context_canceled),
},
"recentFailovers": visible_failovers[-8:],
"recentForwardFailures": visible_forward_failures[-8:],
"recentFinalErrors": visible_final_errors[-8:],
"recentSlowFinalErrors": visible_slow_final_errors[-5:],
"recentContextCanceled": visible_context_canceled[-5:],
"recentProbeNoise": probe_noise[-5:],
"logsExitCode": proc.returncode,
"logsStderr": text(proc.stderr, 1000),
"valuesPrinted": False,
}
def validate_gateway_responses(api_key):
request_id = "unidesk-codex-pool-validate-" + str(int(time.time() * 1000))
payload = {
"model": RESPONSES_SMOKE_MODEL,
"input": "Reply exactly: unidesk-sub2api-validate-ok",
"stream": False,
"store": False,
"max_output_tokens": 32,
}
body = json.dumps(payload, separators=(",", ":")).encode("utf-8")
script = r'''
set -eu
token="$1"
request_id="$2"
tmp="$(mktemp)"
trap 'rm -f "$tmp"' EXIT
cat > "$tmp"
curl -sS -w '\\n__HTTP_CODE__:%{http_code}' -X POST \
-H "Authorization: Bearer $token" \
-H 'Content-Type: application/json' \
-H "X-Request-ID: $request_id" \
-H "OpenAI-Client-Request-ID: $request_id" \
--data-binary @"$tmp" \
http://127.0.0.1:8080/v1/responses
'''
started = time.time()
proc = run([
"kubectl", "-n", NAMESPACE, "exec", "-i", APP_POD,
"--", "sh", "-c", script, "sh", api_key, request_id,
], body)
resp = parse_curl_output(proc)
evidence = request_log_evidence(request_id)
parsed = resp.get("json")
failover_count = len(evidence.get("failovers") or [])
return {
"ok": resp.get("ok"),
"degraded": failover_count > 0,
"outcome": "succeeded-with-failover" if resp.get("ok") and failover_count > 0 else ("succeeded" if resp.get("ok") else "failed"),
"httpStatus": resp.get("httpStatus"),
"transportExitCode": resp.get("transportExitCode"),
"method": "POST /v1/responses",
"model": RESPONSES_SMOKE_MODEL,
"requestId": request_id,
"durationMs": int((time.time() - started) * 1000),
"outputTextPreview": response_output_preview(parsed),
"bodyPreview": "" if resp.get("ok") else text(resp.get("body", ""), 800),
"stderr": resp.get("stderr", ""),
"evidence": evidence,
"valuesPrinted": False,
}
def bool_value(value):
if isinstance(value, bool):
return value
if isinstance(value, str):
if value.lower() == "true":
return True
if value.lower() == "false":
return False
return False
def normalize_temp_unschedulable_credentials(credentials):
if not isinstance(credentials, dict):
credentials = {}
enabled = bool_value(credentials.get("temp_unschedulable_enabled"))
raw_rules = credentials.get("temp_unschedulable_rules")
if isinstance(raw_rules, str):
try:
raw_rules = json.loads(raw_rules)
except json.JSONDecodeError:
raw_rules = []
rules = []
if isinstance(raw_rules, list):
for rule in raw_rules:
if not isinstance(rule, dict):
continue
error_code = rule.get("error_code", rule.get("statusCode"))
duration_minutes = rule.get("duration_minutes", rule.get("durationMinutes"))
keywords = rule.get("keywords")
if not isinstance(error_code, int) or not isinstance(duration_minutes, int) or not isinstance(keywords, list):
continue
clean_keywords = [item for item in keywords if isinstance(item, str) and item.strip()]
if not clean_keywords:
continue
description = rule.get("description") if isinstance(rule.get("description"), str) else ""
rules.append({
"error_code": error_code,
"keywords": clean_keywords,
"duration_minutes": duration_minutes,
"description": description,
})
return {
"enabled": enabled,
"rules": rules,
}
def summarize_temp_unschedulable_rules(rules):
return [{
"errorCode": rule.get("error_code"),
"durationMinutes": rule.get("duration_minutes"),
"keywordCount": len(rule.get("keywords") or []),
"keywords": rule.get("keywords") or [],
"hasDescription": bool(rule.get("description")),
} for rule in rules]
def success_body_reclassification_requirement():
for name in sorted(EXPECTED_ACCOUNT_TEMP_UNSCHEDULABLE):
expected = normalize_temp_unschedulable_credentials(EXPECTED_ACCOUNT_TEMP_UNSCHEDULABLE[name])
if expected["enabled"] is not True:
continue
for rule in expected["rules"]:
error_code = rule.get("error_code")
keywords = rule.get("keywords") or []
if isinstance(error_code, int) and 200 <= error_code < 300 and keywords:
return {
"required": True,
"sourceAccountName": name,
"statusCode": error_code,
"keywords": keywords,
"representativeKeyword": keywords[0],
"durationMinutes": rule.get("duration_minutes"),
}
return {
"required": False,
"sourceAccountName": None,
"statusCode": None,
"keywords": [],
"representativeKeyword": None,
"durationMinutes": None,
}
def model_routing_400_failover_requirement():
preferred = ["invalid_encrypted_content", "encrypted content", "could not be verified", "could not be decrypted", "暂不支持", "可用模型", "unsupported model", "model not supported", "does not support", "not supported", "model_not_found", "no available channel for model"]
for name in sorted(EXPECTED_ACCOUNT_TEMP_UNSCHEDULABLE):
expected = normalize_temp_unschedulable_credentials(EXPECTED_ACCOUNT_TEMP_UNSCHEDULABLE[name])
if expected["enabled"] is not True:
continue
for rule in expected["rules"]:
error_code = rule.get("error_code")
keywords = rule.get("keywords") or []
if error_code != 400 or not keywords:
continue
representative_keyword = next((item for item in preferred if item in keywords), keywords[0])
return {
"required": True,
"sourceAccountName": name,
"statusCode": error_code,
"keywords": keywords,
"representativeKeyword": representative_keyword,
"durationMinutes": rule.get("duration_minutes"),
}
return {
"required": False,
"sourceAccountName": None,
"statusCode": None,
"keywords": [],
"representativeKeyword": None,
"durationMinutes": None,
}
def delete_probe_resource(token, method, path, label):
if not path:
return {"label": label, "ok": True, "skipped": True}
resp = curl_api(method, path, bearer=token)
ok = resp.get("ok") is True or resp.get("httpStatus") in (404, 410)
return {
"label": label,
"ok": ok,
"method": method,
"path": path,
"httpStatus": resp.get("httpStatus"),
"transportExitCode": resp.get("transportExitCode"),
"bodyPreview": "" if ok else text(resp.get("body", ""), 500),
"valuesPrinted": False,
}
def validate_runtime_capabilities(token):
success_body = success_body_reclassification_requirement()
model_routing_400 = model_routing_400_failover_requirement()
return {
"ok": success_body.get("required") is not True and model_routing_400.get("required") is True,
"runtimeImage": app_pod_runtime_image(),
"successBodyReclassification": {
"ok": success_body.get("required") is not True,
"required": success_body.get("required"),
"outcome": "not-required-by-yaml" if success_body.get("required") is not True else "requires-source-review-and-real-traffic-evidence",
"requirement": success_body,
"valuesPrinted": False,
},
"modelRouting400Failover": {
"ok": model_routing_400.get("required") is True,
"required": model_routing_400.get("required"),
"outcome": "yaml-rules-present-runtime-observed-by-real-traffic",
"requirement": model_routing_400,
"evidence": "Use Sub2API source review plus validation.gatewayResponsesRecent and Artificer/real request ids for runtime proof; default validate does not create mock upstreams or temporary failover accounts.",
"valuesPrinted": False,
},
"valuesPrinted": False,
}
def app_pod_runtime_image():
try:
pod = kube_json(["-n", NAMESPACE, "get", "pod", APP_POD], f"pod/{APP_POD}")
spec_containers = ((pod.get("spec") or {}).get("containers") or []) if isinstance(pod, dict) else []
status_containers = ((pod.get("status") or {}).get("containerStatuses") or []) if isinstance(pod, dict) else []
spec = next((item for item in spec_containers if item.get("name") == "sub2api"), spec_containers[0] if spec_containers else {})
status = next((item for item in status_containers if item.get("name") == "sub2api"), status_containers[0] if status_containers else {})
return {
"pod": APP_POD,
"image": spec.get("image"),
"imageID": status.get("imageID"),
"ready": status.get("ready"),
"restartCount": status.get("restartCount"),
}
except Exception as exc:
return {"pod": APP_POD, "error": str(exc)}
def get_account_detail(token, account):
account_id = account.get("id") if isinstance(account, dict) else None
if account_id is None:
return account
data = ensure_success(curl_api("GET", f"/api/v1/admin/accounts/{account_id}", bearer=token), f"get account {account.get('name')}")
return data if isinstance(data, dict) else account
def account_temp_unschedulable_status(token):
accounts = list_accounts(token)
by_name = {item.get("name"): item for item in accounts if isinstance(item.get("name"), str)}
items = []
missing = []
mismatched = []
enabled_names = []
for name in sorted(EXPECTED_ACCOUNT_TEMP_UNSCHEDULABLE):
expected = normalize_temp_unschedulable_credentials(EXPECTED_ACCOUNT_TEMP_UNSCHEDULABLE[name])
account = by_name.get(name)
if account is None:
missing.append(name)
items.append({
"accountName": name,
"accountId": None,
"expectedEnabled": expected["enabled"],
"runtimeEnabled": None,
"expectedRuleCount": len(expected["rules"]),
"runtimeRuleCount": None,
"ok": False,
})
continue
detail = get_account_detail(token, account)
credentials = detail.get("credentials") if isinstance(detail.get("credentials"), dict) else {}
runtime = normalize_temp_unschedulable_credentials(credentials)
temp_until = detail.get("temp_unschedulable_until") or detail.get("tempUnschedulableUntil")
temp_reason = detail.get("temp_unschedulable_reason") or detail.get("tempUnschedulableReason") or ""
ok = runtime == expected
if expected["enabled"]:
enabled_names.append(name)
if not ok:
mismatched.append(name)
items.append({
"accountName": name,
"accountId": account.get("id"),
"expectedEnabled": expected["enabled"],
"runtimeEnabled": runtime["enabled"],
"expectedRuleCount": len(expected["rules"]),
"runtimeRuleCount": len(runtime["rules"]),
"expectedRules": summarize_temp_unschedulable_rules(expected["rules"]),
"runtimeRules": summarize_temp_unschedulable_rules(runtime["rules"]),
"status": account.get("status"),
"schedulable": account.get("schedulable"),
"tempUnschedulableUntil": temp_until,
"tempUnschedulableReasonPreview": text(str(temp_reason), 500) if temp_reason else "",
"tempUnschedulableSet": temp_until is not None or bool(temp_reason),
"ok": ok,
})
return {
"ok": len(missing) == 0 and len(mismatched) == 0,
"desired": len(EXPECTED_ACCOUNT_TEMP_UNSCHEDULABLE),
"enabled": enabled_names,
"missing": missing,
"mismatched": mismatched,
"items": items,
"valuesPrinted": False,
}
def account_capacity_status(token):
accounts = list_accounts(token)
by_name = {item.get("name"): item for item in accounts if isinstance(item.get("name"), str)}
items = []
missing = []
mismatched = []
expected_capacity_total = 0
runtime_concurrency_total = 0
schedulable_runtime_concurrency_total = 0
available_runtime_concurrency_total = 0
temp_unschedulable_runtime_concurrency_total = 0
for name in sorted(EXPECTED_ACCOUNT_CAPACITIES):
expected = int(EXPECTED_ACCOUNT_CAPACITIES[name])
expected_capacity_total += expected
account = by_name.get(name)
if account is None:
missing.append(name)
items.append({
"accountName": name,
"accountId": None,
"expectedCapacity": expected,
"runtimeConcurrency": None,
"ok": False,
})
continue
runtime = account.get("concurrency")
runtime_int = runtime if isinstance(runtime, int) else 0
runtime_concurrency_total += runtime_int
if account.get("schedulable") is True:
runtime_status = account.get("status")
if runtime_status is None or runtime_status == "active":
runtime_status_ok = True
else:
runtime_status_ok = False
if runtime_status_ok:
schedulable_runtime_concurrency_total += runtime_int
temp_until = account.get("temp_unschedulable_until") or account.get("tempUnschedulableUntil")
temp_reason = account.get("temp_unschedulable_reason") or account.get("tempUnschedulableReason")
if temp_until is None and not bool(temp_reason):
available_runtime_concurrency_total += runtime_int
else:
temp_unschedulable_runtime_concurrency_total += runtime_int
ok = runtime == expected
if not ok:
mismatched.append(name)
items.append({
"accountName": name,
"accountId": account.get("id"),
"expectedCapacity": expected,
"runtimeConcurrency": runtime,
"priority": account.get("priority"),
"status": account.get("status"),
"schedulable": account.get("schedulable"),
"ok": ok,
})
return {
"ok": len(missing) == 0 and len(mismatched) == 0,
"defaultAccountCapacity": POOL_DEFAULT_ACCOUNT_CAPACITY,
"desired": len(EXPECTED_ACCOUNT_CAPACITIES),
"totals": {
"expectedCapacityTotal": expected_capacity_total,
"runtimeConcurrencyTotal": runtime_concurrency_total,
"schedulableRuntimeConcurrencyTotal": schedulable_runtime_concurrency_total,
"availableRuntimeConcurrencyTotal": available_runtime_concurrency_total,
"tempUnschedulableRuntimeConcurrencyTotal": temp_unschedulable_runtime_concurrency_total,
"minOwnerConcurrency": MIN_OWNER_CONCURRENCY,
"minOwnerConcurrencySource": MIN_OWNER_CONCURRENCY_SOURCE,
},
"missing": missing,
"mismatched": mismatched,
"items": items,
"valuesPrinted": False,
}
def account_load_factor_status(token):
accounts = list_accounts(token)
by_name = {item.get("name"): item for item in accounts if isinstance(item.get("name"), str)}
items = []
missing = []
mismatched = []
for name in sorted(EXPECTED_ACCOUNT_LOAD_FACTORS):
expected = int(EXPECTED_ACCOUNT_LOAD_FACTORS[name])
account = by_name.get(name)
if account is None:
missing.append(name)
items.append({
"accountName": name,
"accountId": None,
"expectedLoadFactor": expected,
"runtimeLoadFactor": None,
"ok": False,
})
continue
runtime_raw = account.get("load_factor")
if runtime_raw is None:
detail = get_account_detail(token, account)
runtime_raw = detail.get("load_factor") if isinstance(detail, dict) else None
try:
runtime = int(runtime_raw)
except Exception:
runtime = runtime_raw
ok = runtime == expected
if not ok:
mismatched.append(name)
items.append({
"accountName": name,
"accountId": account.get("id"),
"expectedLoadFactor": expected,
"runtimeLoadFactor": runtime,
"priority": account.get("priority"),
"status": account.get("status"),
"schedulable": account.get("schedulable"),
"ok": ok,
})
return {
"ok": len(missing) == 0 and len(mismatched) == 0,
"defaultAccountLoadFactor": POOL_DEFAULT_ACCOUNT_LOAD_FACTOR,
"desired": len(EXPECTED_ACCOUNT_LOAD_FACTORS),
"missing": missing,
"mismatched": mismatched,
"items": items,
"valuesPrinted": False,
}
def account_ws_v2_status(token):
accounts = list_accounts(token)
by_name = {item.get("name"): item for item in accounts if isinstance(item.get("name"), str)}
items = []
missing = []
mismatched = []
enabled_names = []
unschedulable_enabled = []
schedulable_enabled = []
for name in sorted(EXPECTED_ACCOUNT_WS_MODES):
expected_mode = EXPECTED_ACCOUNT_WS_MODES[name]
expected_enabled = expected_mode not in (None, "", "off")
account = by_name.get(name)
if account is None:
missing.append(name)
items.append({
"accountName": name,
"accountId": None,
"expectedMode": expected_mode,
"expectedEnabled": expected_enabled,
"runtimeMode": None,
"runtimeEnabled": None,
"ok": False,
})
continue
extra = account.get("extra") if isinstance(account.get("extra"), dict) else {}
runtime_mode = extra.get("openai_apikey_responses_websockets_v2_mode")
runtime_enabled = extra.get("openai_apikey_responses_websockets_v2_enabled")
schedulable = account.get("schedulable")
if expected_mode == "off":
ok = runtime_mode == "off" and runtime_enabled is False
elif expected_mode is None:
ok = runtime_mode in (None, "", "off") and runtime_enabled in (None, False)
else:
ok = runtime_mode == expected_mode and runtime_enabled is True
if not ok:
mismatched.append(name)
if expected_enabled:
enabled_names.append(name)
if schedulable is True:
schedulable_enabled.append(name)
else:
unschedulable_enabled.append(name)
items.append({
"accountName": name,
"accountId": account.get("id"),
"expectedMode": expected_mode,
"expectedEnabled": expected_enabled,
"runtimeMode": runtime_mode,
"runtimeEnabled": runtime_enabled,
"status": account.get("status"),
"schedulable": schedulable,
"ok": ok and ((not expected_enabled) or schedulable is True),
})
availability_ok = len(enabled_names) == 0 or len(schedulable_enabled) > 0
return {
"ok": len(missing) == 0 and len(mismatched) == 0 and availability_ok,
"desired": len(EXPECTED_ACCOUNT_WS_MODES),
"enabled": enabled_names,
"schedulableEnabled": schedulable_enabled,
"unschedulableEnabled": unschedulable_enabled,
"missing": missing,
"mismatched": mismatched,
"items": items,
"valuesPrinted": False,
}
def api_key_preview(api_key):
if len(api_key) <= 14:
return "***"
return api_key[:10] + "..." + api_key[-4:]
def run_sync():
global MANUAL_ACCOUNT_PROTECTIONS
payload = json.loads(base64.b64decode(PAYLOAD_B64).decode("utf-8"))
manual_accounts_payload = payload.get("manualAccounts") if isinstance(payload.get("manualAccounts"), dict) else {}
resolved_manual_protections = manual_accounts_payload.get("protected")
if not isinstance(resolved_manual_protections, list):
raise RuntimeError("sync payload has no manualAccounts.protected binding plan")
MANUAL_ACCOUNT_PROTECTIONS = resolved_manual_protections
profiles = payload.get("profiles") or []
prune_removed = bool(payload.get("pruneRemoved"))
sentinel_payload = payload.get("sentinel") if isinstance(payload.get("sentinel"), dict) else {}
if not profiles:
raise RuntimeError("sync payload has no profiles")
admin_email, token, admin_compliance = login()
group, group_action = ensure_group(token)
group_id = group.get("id") if isinstance(group, dict) else None
if group_id is None:
raise RuntimeError("pool group id missing after ensure")
existing_accounts = list_accounts(token)
planned_account_results = planned_sentinel_account_results(profiles, existing_accounts)
sentinel_quality_prepare = ensure_sentinel_state_for_sync(planned_account_results, True)
if sentinel_quality_prepare.get("ok") is not True:
raise RuntimeError("prepare sentinel pending probe failed: " + json.dumps(sentinel_quality_prepare, ensure_ascii=False))
protected_frozen_names = active_sentinel_quarantine_names()
account_results, pruned_account_results = ensure_accounts(token, profiles, group_id, prune_removed, protected_frozen_names, existing_accounts)
manual_account_proxy_bindings = ensure_manual_account_proxy_bindings(token)
manual_account_group_bindings = ensure_manual_account_group_bindings(token, group_id)
manual_account_protections = manual_account_protection_status(token, group_id)
capacity_status = account_capacity_status(token)
load_factor_status = account_load_factor_status(token)
ws_v2_status = account_ws_v2_status(token)
temp_unschedulable_status = account_temp_unschedulable_status(token)
api_key, secret_action, secret_apply_stdout = ensure_api_key_secret(group_id)
api_key_result = ensure_sub2api_api_key(token, api_key, group_id)
owner_balance = ensure_pool_owner_balance(token, api_key_result["userId"])
owner_concurrency = ensure_pool_owner_concurrency(token, api_key_result["userId"])
gateway = validate_gateway(api_key)
responses_smoke = validate_gateway_responses(api_key)
compact_evidence = recent_compact_gateway_evidence()
responses_evidence = recent_responses_gateway_evidence()
runtime_capabilities = validate_runtime_capabilities(token)
sentinel = apply_sentinel_manifest(sentinel_payload.get("manifest"))
sentinel_quality = ensure_sentinel_state_for_sync(account_results)
sentinel_reassert = reassert_sentinel_freezes_after_sync(token)
return {
"ok": gateway["ok"] is True and responses_smoke["ok"] is True and owner_concurrency["ok"] is True and capacity_status["ok"] is True and load_factor_status["ok"] is True and ws_v2_status["ok"] is True and temp_unschedulable_status["ok"] is True and manual_account_proxy_bindings.get("ok") is True and manual_account_group_bindings.get("ok") is True and manual_account_protections.get("ok") is True and sentinel.get("ok") is True and sentinel_quality_prepare.get("ok") is True and sentinel_quality.get("ok") is True and sentinel_reassert.get("ok") is True and runtime_capabilities.get("ok") is True,
"degraded": bool(responses_smoke.get("degraded")) or bool(compact_evidence.get("degraded")) or bool(responses_evidence.get("degraded")) or runtime_capabilities.get("ok") is not True,
"mode": "sync",
"namespace": NAMESPACE,
"serviceDns": SERVICE_DNS,
"appPod": APP_POD,
"admin": {"email": admin_email, "tokenPrinted": False, "compliance": admin_compliance},
"pool": {"name": POOL_GROUP_NAME, "id": group_id, "action": group_action, "platform": group.get("platform") if isinstance(group, dict) else "openai"},
"accounts": {
"desired": len(profiles),
"created": sum(1 for item in account_results if item["action"] == "created"),
"updated": sum(1 for item in account_results if item["action"] == "updated"),
"pruned": len(pruned_account_results),
"pruneMode": "explicit" if prune_removed else "disabled-by-default",
"items": account_results,
"prunedItems": pruned_account_results,
"processControl": {"schedulableRestore": "sentinel marker probe only; sync does not restore schedulable for existing accounts", "durableConfig": False},
"valuesPrinted": False,
},
"manualAccounts": {**manual_account_protections, "proxySync": manual_account_proxy_bindings, "groupSync": manual_account_group_bindings},
"capacity": capacity_status,
"loadFactor": load_factor_status,
"webSocketsV2": ws_v2_status,
"tempUnschedulable": temp_unschedulable_status,
"apiKey": {
"name": POOL_API_KEY_NAME,
"secret": f"{NAMESPACE}/{POOL_API_KEY_SECRET_NAME}.{POOL_API_KEY_SECRET_KEY}",
"secretAction": secret_action,
"secretApply": secret_apply_stdout,
"sub2apiAction": api_key_result["action"],
"sub2apiId": api_key_result["id"],
"groupId": api_key_result["groupId"],
"userId": api_key_result["userId"],
"keyPreview": api_key_preview(api_key),
"valuesPrinted": False,
},
"ownerBalance": owner_balance,
"ownerConcurrency": owner_concurrency,
"sentinel": {**sentinel, "qualityGatePrepare": sentinel_quality_prepare, "qualityGate": sentinel_quality, "freezeReassert": sentinel_reassert},
"runtimeCapabilities": runtime_capabilities,
"validation": {"gatewayModels": gateway, "gatewayResponses": responses_smoke, "gatewayResponsesRecent": responses_evidence, "gatewayCompactRecent": compact_evidence},
}
def run_validate():
api_key = decode_secret_value(POOL_API_KEY_SECRET_NAME, POOL_API_KEY_SECRET_KEY)
if not api_key:
raise RuntimeError(f"{POOL_API_KEY_SECRET_NAME}.{POOL_API_KEY_SECRET_KEY} missing")
admin_email, token, admin_compliance = login()
key_item = next((item for item in list_user_keys(token) if item.get("key") == api_key), None)
owner_balance = None
owner_concurrency = None
if key_item is not None and key_item.get("user_id") is not None:
owner_balance = ensure_pool_owner_balance(token, key_item["user_id"])
owner_concurrency = ensure_pool_owner_concurrency(token, key_item["user_id"])
capacity_status = account_capacity_status(token)
load_factor_status = account_load_factor_status(token)
ws_v2_status = account_ws_v2_status(token)
temp_unschedulable_status = account_temp_unschedulable_status(token)
pool_group_id = key_item.get("group_id") if isinstance(key_item, dict) else None
manual_account_protections = manual_account_protection_status(token, pool_group_id)
gateway = validate_gateway(api_key)
responses_smoke = validate_gateway_responses(api_key)
compact_evidence = recent_compact_gateway_evidence()
responses_evidence = recent_responses_gateway_evidence()
runtime_capabilities = validate_runtime_capabilities(token)
sentinel = sentinel_runtime_status()
return {
"ok": gateway["ok"] is True and responses_smoke["ok"] is True and (owner_concurrency is None or owner_concurrency["ok"] is True) and capacity_status["ok"] is True and load_factor_status["ok"] is True and ws_v2_status["ok"] is True and temp_unschedulable_status["ok"] is True and manual_account_protections.get("ok") is True and sentinel.get("ok") is True and runtime_capabilities.get("ok") is True,
"degraded": bool(responses_smoke.get("degraded")) or bool(compact_evidence.get("degraded")) or bool(responses_evidence.get("degraded")) or runtime_capabilities.get("ok") is not True,
"mode": "validate",
"namespace": NAMESPACE,
"serviceDns": SERVICE_DNS,
"appPod": APP_POD,
"admin": {"email": admin_email, "tokenPrinted": False, "compliance": admin_compliance},
"apiKey": {
"secret": f"{NAMESPACE}/{POOL_API_KEY_SECRET_NAME}.{POOL_API_KEY_SECRET_KEY}",
"sub2apiId": key_item.get("id") if isinstance(key_item, dict) else None,
"userId": key_item.get("user_id") if isinstance(key_item, dict) else None,
"groupId": key_item.get("group_id") if isinstance(key_item, dict) else None,
"keyPreview": api_key_preview(api_key),
"valuesPrinted": False,
},
"ownerBalance": owner_balance,
"ownerConcurrency": owner_concurrency,
"capacity": capacity_status,
"loadFactor": load_factor_status,
"webSocketsV2": ws_v2_status,
"tempUnschedulable": temp_unschedulable_status,
"manualAccounts": manual_account_protections,
"sentinel": sentinel,
"runtimeCapabilities": runtime_capabilities,
"validation": {"gatewayModels": gateway, "gatewayResponses": responses_smoke, "gatewayResponsesRecent": responses_evidence, "gatewayCompactRecent": compact_evidence},
}
def parse_log_line(line):
json_start = line.find("{")
if json_start < 0:
return None
prefix = line[:json_start].rstrip()
try:
item = json.loads(line[json_start:])
except Exception:
return None
if not isinstance(item, dict):
return None
at = None
parts = prefix.split()
if parts:
at = parts[0]
message = ""
if len(parts) >= 4:
message = " ".join(parts[3:])
elif len(parts) >= 3:
message = parts[2]
elif len(parts) >= 1:
message = parts[-1]
item["_at"] = at
item["_message"] = message
item["_line"] = line
return item
def log_time_epoch(item):
at = item.get("_at") if isinstance(item, dict) else None
if not isinstance(at, str) or not at:
return None
try:
return datetime.strptime(at, "%Y-%m-%dT%H:%M:%S.%f%z").timestamp()
except Exception:
try:
return datetime.fromisoformat(at.replace("Z", "+00:00")).timestamp()
except Exception:
return None
def event_base(item, account_names_by_id):
account_id = item.get("account_id")
if isinstance(account_id, str) and account_id.isdigit():
account_id = int(account_id)
account_name = account_names_by_id.get(account_id)
return {
"at": item.get("_at"),
"message": item.get("_message"),
"requestId": item.get("request_id"),
"clientRequestId": item.get("client_request_id"),
"path": item.get("path"),
"method": item.get("method"),
"model": item.get("model"),
"accountId": account_id,
"accountName": account_name,
"statusCode": item.get("status_code"),
"upstreamStatus": item.get("upstream_status"),
"latencyMs": item.get("latency_ms"),
}
def classify_trace_event(item, account_names_by_id):
message = str(item.get("_message") or "")
event = event_base(item, account_names_by_id)
if "content_moderation.gateway_check_start" in message:
event.update({
"type": "request-start",
"stream": item.get("stream"),
"bodyBytes": item.get("body_bytes"),
"groupId": item.get("group_id"),
"groupName": item.get("group_name"),
"apiKeyName": item.get("api_key_name"),
})
elif "content_moderation.gateway_check_done" in message:
event.update({
"type": "gateway-check",
"allowed": item.get("allowed"),
"blocked": item.get("blocked"),
"action": item.get("action"),
})
elif "openai.upstream_failover_switching" in message:
event.update({
"type": "failover",
"switchCount": item.get("switch_count"),
"maxSwitches": item.get("max_switches"),
})
elif "openai.account_select_failed" in message:
event.update({
"type": "select-failed",
"error": item.get("error"),
"excludedAccountCount": item.get("excluded_account_count"),
})
elif "account_upstream_error" in message:
event.update({
"type": "upstream-error",
"error": item.get("error"),
})
elif "account_temp_unschedulable" in message:
event.update({
"type": "temp-unschedulable",
"until": item.get("until") or item.get("temp_unschedulable_until"),
"ruleIndex": item.get("rule_index"),
"matchedKeyword": item.get("matched_keyword"),
"reason": item.get("reason") or item.get("error"),
})
elif "http request completed" in message:
event.update({
"type": "final",
"clientIp": item.get("client_ip"),
"protocol": item.get("protocol"),
"platform": item.get("platform"),
"completedAt": item.get("completed_at"),
})
elif "admin account schedulable updated" in message or "account schedulable updated" in message or "/schedulable" in str(item.get("path") or ""):
event.update({
"type": "admin-schedulable",
"schedulable": item.get("schedulable"),
})
else:
event.update({"type": "other"})
return event
def with_trace_phase(event, first_epoch, last_epoch):
epoch = None
at = event.get("at") if isinstance(event, dict) else None
if isinstance(at, str) and at:
epoch = log_time_epoch({"_at": at})
if epoch is None or first_epoch is None:
phase = "unknown"
elif epoch < first_epoch:
phase = "before-request"
elif last_epoch is not None and epoch > last_epoch:
phase = "after-request"
else:
phase = "during-request"
event["phase"] = phase
return event
def account_snapshot_from_runtime(token):
try:
accounts = list_accounts(token)
except Exception as exc:
return [], {"error": str(exc)}
rows = []
for item in accounts:
if not isinstance(item, dict):
continue
rows.append({
"accountId": item.get("id"),
"accountName": item.get("name"),
"schedulable": item.get("schedulable"),
"status": item.get("status"),
"concurrency": item.get("concurrency"),
"priority": item.get("priority"),
"tempUnschedulableUntil": item.get("temp_unschedulable_until") or item.get("tempUnschedulableUntil"),
"tempUnschedulableSet": (item.get("temp_unschedulable_until") or item.get("tempUnschedulableUntil")) is not None or bool(item.get("temp_unschedulable_reason") or item.get("tempUnschedulableReason")),
})
rows.sort(key=lambda row: (str(row.get("accountName") or ""), int(row.get("accountId") or 0)))
return rows, None
def trace_reason(events, final_event):
failovers = [item for item in events if item.get("type") == "failover"]
select_failures = [item for item in events if item.get("type") == "select-failed"]
upstream_errors = [item for item in events if item.get("type") == "upstream-error"]
if failover_budget_exhausted(failovers, final_event):
return "failover-budget-exhausted"
if failovers and select_failures:
return "failover-attempted-no-candidate"
if failovers:
return "failover-attempted"
if select_failures:
return "account-select-failed"
if upstream_errors:
return "upstream-error"
if isinstance(final_event, dict) and isinstance(final_event.get("statusCode"), int) and final_event.get("statusCode") >= 400:
return "final-http-error"
if isinstance(final_event, dict) and isinstance(final_event.get("statusCode"), int):
return "completed"
return "unknown"
def failover_budget_exhausted(failovers, final_event):
if not failovers or not isinstance(final_event, dict):
return False
last = failovers[-1]
switch_count = last.get("switchCount")
max_switches = last.get("maxSwitches")
final_status = final_event.get("statusCode")
return (
isinstance(switch_count, int)
and isinstance(max_switches, int)
and max_switches > 0
and switch_count >= max_switches
and isinstance(final_status, int)
and final_status >= 500
)
def trace_untried_schedulable_accounts(failovers, final_event, account_snapshot):
if not failover_budget_exhausted(failovers, final_event):
return []
tried = set()
for item in failovers:
account_id = item.get("accountId")
if isinstance(account_id, int):
tried.add(account_id)
final_account = final_event.get("accountId") if isinstance(final_event, dict) else None
if isinstance(final_account, int):
tried.add(final_account)
result = []
for item in account_snapshot:
account_id = item.get("accountId")
if not isinstance(account_id, int) or account_id in tried:
continue
if item.get("schedulable") is True and item.get("status") == "active" and item.get("tempUnschedulableSet") is not True:
result.append({
"accountId": account_id,
"accountName": item.get("accountName"),
"priority": item.get("priority"),
"concurrency": item.get("concurrency"),
})
return result
def run_trace():
payload = json.loads(base64.b64decode(PAYLOAD_B64).decode("utf-8")) if PAYLOAD_B64 else {}
request_id = payload.get("requestId")
since = payload.get("since") or "24h"
tail = int(payload.get("tail") or 20000)
context_seconds = int(payload.get("contextSeconds") or 300)
show_lines = bool(payload.get("showLines"))
if not isinstance(request_id, str) or not request_id:
raise RuntimeError("trace payload missing requestId")
admin_email, token, admin_compliance = login()
account_snapshot, account_snapshot_error = account_snapshot_from_runtime(token)
account_names_by_id = {}
for row in account_snapshot:
account_id = row.get("accountId")
if isinstance(account_id, str) and account_id.isdigit():
account_id = int(account_id)
if isinstance(account_id, int) and isinstance(row.get("accountName"), str):
account_names_by_id[account_id] = row.get("accountName")
proc = kubectl(["-n", NAMESPACE, "logs", "deployment/sub2api", f"--since={since}", f"--tail={tail}"])
stdout = proc.stdout.decode("utf-8", errors="replace")
parsed_lines = []
matched = []
for line in stdout.splitlines():
parsed = parse_log_line(line)
if parsed is None:
continue
parsed_lines.append(parsed)
if request_id in line:
matched.append(parsed)
first_epoch = None
last_epoch = None
for item in matched:
epoch = log_time_epoch(item)
if epoch is None:
continue
first_epoch = epoch if first_epoch is None else min(first_epoch, epoch)
last_epoch = epoch if last_epoch is None else max(last_epoch, epoch)
window_lines = []
if first_epoch is not None:
start_epoch = first_epoch - context_seconds
end_epoch = (last_epoch if last_epoch is not None else first_epoch) + context_seconds
for item in parsed_lines:
epoch = log_time_epoch(item)
if epoch is not None and start_epoch <= epoch <= end_epoch:
window_lines.append(item)
else:
window_lines = matched
events = [classify_trace_event(item, account_names_by_id) for item in matched]
request_start = next((item for item in events if item.get("type") == "request-start"), None)
final_event = next((item for item in reversed(events) if item.get("type") == "final"), None)
failovers = [item for item in events if item.get("type") == "failover"]
select_failures = [item for item in events if item.get("type") == "select-failed"]
upstream_errors = [item for item in events if item.get("type") == "upstream-error"]
temp_unsched = [with_trace_phase(classify_trace_event(item, account_names_by_id), first_epoch, last_epoch) for item in window_lines if "account_temp_unschedulable" in str(item.get("_message") or "")]
admin_sched = [with_trace_phase(classify_trace_event(item, account_names_by_id), first_epoch, last_epoch) for item in window_lines if ("schedulable" in str(item.get("_message") or "") or "/schedulable" in str(item.get("path") or ""))]
window_events = [classify_trace_event(item, account_names_by_id) for item in window_lines]
final_errors = [item for item in window_events if item.get("type") == "final" and isinstance(item.get("statusCode"), int) and item.get("statusCode") >= 400]
window_failovers = [item for item in window_events if item.get("type") == "failover"]
window_select_failures = [item for item in window_events if item.get("type") == "select-failed"]
untried_schedulable_accounts = trace_untried_schedulable_accounts(failovers, final_event or {}, account_snapshot)
reason = trace_reason(events, final_event)
if not matched:
outcome = "not-found"
elif isinstance(final_event, dict) and isinstance(final_event.get("statusCode"), int) and final_event.get("statusCode") < 400:
outcome = "succeeded"
elif isinstance(final_event, dict):
outcome = "failed"
else:
outcome = "incomplete"
return {
"ok": proc.returncode == 0 and len(matched) > 0,
"mode": "trace",
"namespace": NAMESPACE,
"serviceDns": SERVICE_DNS,
"appPod": APP_POD,
"admin": {"email": admin_email, "tokenPrinted": False, "compliance": admin_compliance},
"requestId": request_id,
"summary": {
"outcome": outcome,
"reason": reason,
"eventCount": len(events),
"matchedLineCount": len(matched),
"firstAt": events[0].get("at") if events else None,
"lastAt": events[-1].get("at") if events else None,
},
"window": {
"since": since,
"tail": tail,
"beforeSeconds": context_seconds,
"afterSeconds": context_seconds,
"lineCount": len(window_lines),
},
"request": request_start or {},
"final": final_event or {},
"events": events,
"failovers": failovers,
"selectFailures": select_failures,
"upstreamErrors": upstream_errors,
"tempUnschedulable": temp_unsched,
"adminSchedulable": admin_sched[-20:],
"windowStats": {
"matchedLines": len(matched),
"eventCount": len(window_events),
"finalErrorCount": len(final_errors),
"failoverCount": len(window_failovers),
"selectFailedCount": len(window_select_failures),
"tempUnschedulableCount": len(temp_unsched),
"adminSchedulableCount": len(admin_sched),
},
"diagnostics": {
"failoverBudgetExhausted": failover_budget_exhausted(failovers, final_event or {}),
"untriedSchedulableAccounts": untried_schedulable_accounts,
},
"accountSnapshot": account_snapshot,
"accountSnapshotError": account_snapshot_error,
"rawLines": [{"line": item.get("_line")} for item in matched[-30:]] if show_lines else [],
"showLines": show_lines,
"logs": {
"exitCode": proc.returncode,
"stderrTail": text(proc.stderr, 1000),
"stdoutLineCount": len(stdout.splitlines()),
},
"valuesPrinted": False,
}
def parse_embedded_json(stdout):
if not isinstance(stdout, str) or not stdout.strip():
return None
start = stdout.find("{")
end = stdout.rfind("}")
if start < 0 or end <= start:
return None
try:
return json.loads(stdout[start:end + 1])
except Exception:
return None
def inject_env(template, name, value):
spec = template.setdefault("spec", {})
containers = spec.setdefault("containers", [])
if not containers:
raise RuntimeError("sentinel job template has no containers")
container = containers[0]
env = container.setdefault("env", [])
env[:] = [item for item in env if not (isinstance(item, dict) and item.get("name") == name)]
env.append({"name": name, "value": value})
def sentinel_probe_job_manifest(accounts):
cronjob_name = SENTINEL_CONFIG.get("cronJobName")
if not isinstance(cronjob_name, str) or not cronjob_name:
raise RuntimeError("sentinel cronJobName missing from config")
cronjob = kube_json(["-n", NAMESPACE, "get", "cronjob", cronjob_name], f"cronjob/{cronjob_name}")
job_template = ((cronjob.get("spec") or {}).get("jobTemplate") or {}) if isinstance(cronjob, dict) else {}
job_spec = copy_json(job_template.get("spec") or {})
if not isinstance(job_spec, dict) or not job_spec:
raise RuntimeError("sentinel CronJob jobTemplate.spec missing")
template = job_spec.get("template")
if not isinstance(template, dict):
raise RuntimeError("sentinel CronJob jobTemplate.spec.template missing")
inject_env(template, "SENTINEL_ACCOUNT_NAMES", ",".join(accounts))
job_spec["ttlSecondsAfterFinished"] = int(job_spec.get("ttlSecondsAfterFinished") or 3600)
suffix = str(int(time.time()))[-8:] + "-" + "".join(secrets.choice(string.ascii_lowercase + string.digits) for _ in range(5))
job_name = ("sub2api-sentinel-probe-" + suffix)[:63]
return job_name, {
"apiVersion": "batch/v1",
"kind": "Job",
"metadata": {
"name": job_name,
"namespace": NAMESPACE,
"labels": {
"app.kubernetes.io/name": cronjob_name,
"app.kubernetes.io/part-of": "platform-infra",
"app.kubernetes.io/managed-by": "unidesk",
"unidesk.ai/job-purpose": "sub2api-account-sentinel-manual-probe",
},
},
"spec": job_spec,
}
def copy_json(value):
return json.loads(json.dumps(value))
def job_condition(job, cond_type):
for item in ((job.get("status") or {}).get("conditions") or []):
if item.get("type") == cond_type and item.get("status") == "True":
return item
return None
def wait_sentinel_probe_job(job_name, timeout_seconds):
deadline = time.time() + timeout_seconds
latest = None
while time.time() < deadline:
latest, err = safe_kube_json(["-n", NAMESPACE, "get", "job", job_name], f"job/{job_name}")
if isinstance(latest, dict):
complete = job_condition(latest, "Complete")
failed = job_condition(latest, "Failed")
if complete is not None:
return "succeeded", latest, complete
if failed is not None:
return "failed", latest, failed
time.sleep(2)
return "timeout", latest, None
def job_logs(job_name):
proc = kubectl(["-n", NAMESPACE, "logs", f"job/{job_name}", "--tail=4000"])
return {
"exitCode": proc.returncode,
"stdout": proc.stdout.decode("utf-8", errors="replace"),
"stderr": text(proc.stderr, 4000),
}
def run_sentinel_probe():
payload = json.loads(base64.b64decode(PAYLOAD_B64).decode("utf-8")) if PAYLOAD_B64 else {}
accounts = payload.get("accounts") if isinstance(payload, dict) else None
if not isinstance(accounts, list) or not accounts or not all(isinstance(item, str) and item for item in accounts):
raise RuntimeError("sentinel-probe payload requires non-empty accounts")
job_name, manifest = sentinel_probe_job_manifest(accounts)
proc = kubectl(["apply", "--server-side", "--force-conflicts", f"--field-manager={FIELD_MANAGER}", "-f", "-"], json.dumps(manifest))
if proc.returncode != 0:
raise RuntimeError(f"apply sentinel probe job failed: {text(proc.stderr, 2000)}")
timeout_seconds = max(300, int(SENTINEL_CONFIG.get("probe", {}).get("timeoutSeconds") or 30) + 240)
status, job, condition = wait_sentinel_probe_job(job_name, timeout_seconds)
logs = job_logs(job_name)
parsed = parse_embedded_json(logs.get("stdout") or "")
results = parsed.get("results") if isinstance(parsed, dict) and isinstance(parsed.get("results"), list) else []
requested = set(accounts)
measured = {item.get("accountName") for item in results if isinstance(item, dict)}
missing = sorted(name for name in requested if name not in measured)
marker_ok = len(missing) == 0 and all(isinstance(item, dict) and item.get("accountName") in requested and item.get("markerMatched") is True for item in results if isinstance(item, dict) and item.get("accountName") in requested)
job_ok = status == "succeeded" and isinstance(parsed, dict) and parsed.get("ok") is True
return {
"ok": job_ok and marker_ok,
"jobExecutionOk": job_ok,
"markerOk": marker_ok,
"mode": "sentinel-probe",
"namespace": NAMESPACE,
"requestedAccounts": accounts,
"missingAccounts": missing,
"job": {
"name": job_name,
"status": status,
"condition": condition,
"timeoutSeconds": timeout_seconds,
"applyStdoutTail": text(proc.stdout, 1200),
"logsExitCode": logs.get("exitCode"),
"logsStderrTail": logs.get("stderr"),
},
"probe": parsed,
"sentinelState": sentinel_state_summary(),
"valuesPrinted": False,
}
def run_cleanup_probes():
admin_email, token, admin_compliance = login()
cleanup = cleanup_probe_resources(token)
return {
"ok": cleanup.get("ok") is True,
"mode": "cleanup-probes",
"namespace": NAMESPACE,
"serviceDns": SERVICE_DNS,
"appPod": APP_POD,
"admin": {"email": admin_email, "tokenPrinted": False, "compliance": admin_compliance},
"cleanup": cleanup,
"valuesPrinted": False,
}
try:
if MODE == "sync":
result = run_sync()
elif MODE == "trace":
result = run_trace()
elif MODE == "cleanup-probes":
result = run_cleanup_probes()
elif MODE == "sentinel-probe":
result = run_sentinel_probe()
else:
result = run_validate()
except Exception as exc:
result = {
"ok": False,
"mode": MODE,
"namespace": NAMESPACE,
"serviceDns": SERVICE_DNS,
"appPod": globals().get("APP_POD"),
"error": str(exc),
"valuesPrinted": False,
}
print(json.dumps(result, ensure_ascii=False, indent=2))
sys.exit(0 if result.get("ok") else 1)
PY
`;
}