Resource Exhaustion
159 incident problems that show up as “Resource Exhaustion”.
먼저 읽을 가이드
추천 문제
All problems (159)
K8S-060HPA and PodDisruptionBudget combine to deadlock a low-replica rolloutAutoscaling and disruption control each look reasonable alone, but together they prevent the deployment from making forward progress during an update.KubernetesAdvanced28 minProLINUX-146An rsyslog ruleset drops duplicate messages aggressively, and a brute-force pattern disappearsThe host is quieter, but the deduplication logic removed meaningful repetition that analysts rely on.LinuxIntermediate15 minProLINUX-092Kernel parameter is tuned in sysctl.d but an earlier file winsThe desired setting exists on disk, yet the host still uses another value because an earlier or later file in the sysctl load order overrides it.LinuxAdvanced15 minProCICD-134A dependency proxy serves a cached package manifest after the source package was revoked, and downstream builds keep resolving the unsafe versionUpstream fixed the issue, but the local acceleration layer preserved the vulnerable metadata view.CI/CDIntermediate16 minProLINUX-152A thin pool autoextend threshold is correct, but the monitoring alert watches data percent while metadata is the first resource to exhaustOperators think storage headroom is fine until metadata starvation freezes writes.LinuxAdvanced16 minProLINUX-192Noise suppression removes the repeated signal analysts rely on for brute-force detection during a failover rehearsalThe logs are cleaner and the incident pattern disappears with the noise. Normal traffic masked the issue until the standby or alternate path became active under rehearsal conditions.LinuxIntermediate16 minProK8S-369A drained node returns to service (Resource Exhaustion)A drained node returns to service (Resource Exhaustion) focuses on kubernetes-workload-reliability and asks the reader to isolate Resource Exhaustion. 실무에서는 resource-exhaustion 증상만 보고 Pod 하나에 매달리지 말고 이벤트, 이전 로그, Service/Endpoint, 최근 배포 변경을 한 번에 묶어 보는 편이 오진을 줄입니다.KubernetesAdvanced17 minProLINUX-149A kdump target path exists, but the crash kernel reserves too little memory after a kernel update and dumps truncate silentlyCrash capture remains enabled in config, yet one sizing assumption no longer holds for the new kernel footprint.LinuxAdvanced17 minProLINUX-369A storage cleanup frees file blocks while an abandoned overlay or container layer still holds the inode pressure that triggered the incident during a staged decommissionCapacity charts improve and the inode shortage remains. The service still works through the primary path, but one dependency only fails when the old component is finally drained away.LinuxAdvanced17 minProLINUX-258A thin snapshot is mounted read-only while one housekeeping timer still tries to trim and purge inside it during a failover rehearsalThe maintenance timer keeps running against a snapshot that no longer accepts the same write behavior. Normal traffic masked the issue until the standby or alternate path became active under rehearsal conditions.LinuxIntermediate17 minProK8S-148An HPA based on external queue depth scales up correctly, but scale-down never happensThe autoscaler is healthy, yet one observability component keeps a historical answer longer than expected.KubernetesAdvanced17 minProLINUX-157An mdadm write-intent bitmap survives the array move, but sector size assumptions changed and recovery performance collapses on the new hostThe array is valid, yet one metadata optimization no longer matches the hardware characteristics.LinuxAdvanced17 minProLINUX-125An rsync inplace copy lands on an XFS reflink-based volume and backup verification later finds silently shared extentsThe transfer finished, but the data isolation assumption behind the backup strategy no longer holds.LinuxAdvanced17 minProLINUX-118cgroup v2 memory.high throttles the workers and the service looks slow instead of obviously brokenLatency climbs without hard OOM events because the workload is being pressured by the soft limit long before memory.max is reached.LinuxAdvanced17 minProK8S-129Topology-aware routing and sticky sessions combine to create a hotspot on one zone after a scale eventNeither feature is individually wrong, but together they pin too much traffic to a shrinking endpoint subset.KubernetesAdvanced17 minProK8S-160A node image upgrade enables cgroup v2, and one Java workload starts misreporting heap limits so the HPA scales on the wrong utilization basisThe app still runs, but runtime memory accounting changed with the node image and distorts autoscaling decisions.KubernetesAdvanced18 minProLINUX-306A thin pool data volume has headroom while metadata snapshots from a backup workflow consume the limiting resource first during a failover rehearsalCapacity planning appears healthy until the hidden metadata consumer wins the race. Normal traffic masked the issue until the standby or alternate path became active under rehearsal conditions.LinuxAdvanced18 minProK8S-204An autoscaler trusts a queue-depth metric that remains stale long after backlog is gone during a failover rehearsalScaling keeps behaving as if yesterday's pressure still exists because freshness is not part of the metric contract. Normal traffic masked the issue until the standby or alternate path became active under rehearsal conditions.KubernetesAdvanced18 minProLINUX-145An LVM cache pool masks a failing SSD, and writeback mode keeps acknowledging writes until the cache metadata finally goes read-onlyStorage looks fast and healthy right up to the point where recovery becomes much harder because the failure mode was buffered.LinuxAdvanced18 minProLINUX-138An mdadm bitmap keeps recovery fast on paper, but a write-mostly member slows resync so much that the array never catches up under loadThe storage survives, yet the degraded state lingers because one tuning decision and one hardware profile conflict.LinuxAdvanced18 minPro