Baoyuantop commented on issue #2734: URL: https://github.com/apache/apisix-ingress-controller/issues/2734#issuecomment-4107741309
Thanks for the detailed logs. From the log timeline, the failure window shows kube-apiserver connection failures, leader election loss, and API discovery failures, followed by persistent `Index with name field:XXX does not exist` errors. Based on the code, AIC checks API resource availability via discovery during startup to decide whether to register field indexes. If the API server is unreachable at that point, index registration gets skipped. Once the cluster recovers, the indexes are never retroactively registered, causing continuous Reconcile failures until restart. To help narrow down the root cause further, could you provide: 1. Whether the ingress-controller pod was recreated during/after failover (and if so, whether the API server was still unstable at that time) 2. Full startup logs showing `api-detection` / `Skipping indexer setup` lines 3. Whether any CRD/APIService availability changed during failover 4. A minimal reproduction procedure if possible -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
