# Copied from /deploy/charts/rook-ceph-cluster/prometheus/, CR header added, and indentation increased on the groups apiVersion: monitoring.coreos.com/v1 kind: PrometheusRule metadata: labels: prometheus: rook-prometheus role: alert-rules name: prometheus-ceph-rules namespace: rook-ceph spec: groups: - name: cluster health rules: - alert: CephHealthError expr: ceph_health_status == 2 for: 5m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.2.1 annotations: summary: Cluster is in the ERROR state description: > The cluster state has been HEALTH_ERROR for more than 5 minutes. Please check "ceph health detail" for more information. - alert: CephHealthWarning expr: ceph_health_status == 1 for: 15m labels: severity: warning type: ceph_default annotations: summary: Cluster is in the WARNING state description: > The cluster state has been HEALTH_WARN for more than 15 minutes. Please check "ceph health detail" for more information. - name: mon rules: - alert: CephMonDownQuorumAtRisk expr: ((ceph_health_detail{name="MON_DOWN"} == 1) * on() (count(ceph_mon_quorum_status == 1) == bool (floor(count(ceph_mon_metadata) / 2) + 1))) == 1 for: 30s labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.3.1 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#mon-down summary: Monitor quorum is at risk description: | {{ $min := query "floor(count(ceph_mon_metadata) / 2) +1" | first | value }}Quorum requires a majority of monitors (x {{ $min }}) to be active Without quorum the cluster will become inoperable, affecting all services and connected clients. The following monitors are down: {{- range query "(ceph_mon_quorum_status == 0) + on(ceph_daemon) group_left(hostname) (ceph_mon_metadata * 0)" }} - {{ .Labels.ceph_daemon }} on {{ .Labels.hostname }} {{- end }} - alert: CephMonDown expr: (count(ceph_mon_quorum_status == 0) <= (count(ceph_mon_metadata) - floor(count(ceph_mon_metadata) / 2) + 1)) for: 30s labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#mon-down summary: One or more monitors down description: | {{ $down := query "count(ceph_mon_quorum_status == 0)" | first | value }}{{ $s := "" }}{{ if gt $down 1.0 }}{{ $s = "s" }}{{ end }}There are {{ $down }} monitor{{ $s }} down. Quorum is still intact, but the loss of an additional monitor will make your cluster inoperable. The following monitors are down: {{- range query "(ceph_mon_quorum_status == 0) + on(ceph_daemon) group_left(hostname) (ceph_mon_metadata * 0)" }} - {{ .Labels.ceph_daemon }} on {{ .Labels.hostname }} {{- end }} - alert: CephMonDiskspaceCritical expr: ceph_health_detail{name="MON_DISK_CRIT"} == 1 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.3.2 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#mon-disk-crit summary: Filesystem space on at least one monitor is critically low description: | The free space available to a monitor's store is critically low. You should increase the space available to the monitor(s). The default directory is /var/lib/ceph/mon-*/data/store.db on traditional deployments, and under /var/lib/rook/mon-*/data/store.db on the mon pod's worker node for Rook. Look for old, rotated versions of *.log and MANIFEST*. Do NOT touch any *.sst files. Also check any other directories under /var/lib/rook and other directories on the same filesystem, often /var/log and /var/tmp are culprits. Your monitor hosts are; {{- range query "ceph_mon_metadata"}} - {{ .Labels.hostname }} {{- end }} - alert: CephMonDiskspaceLow expr: ceph_health_detail{name="MON_DISK_LOW"} == 1 for: 5m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#mon-disk-low summary: Disk space on at least one monitor is approaching full description: | The space available to a monitor's store is approaching full (>70% is the default). You should increase the space available to the monitor(s). The default directory is /var/lib/ceph/mon-*/data/store.db on traditional deployments, and under /var/lib/rook/mon-*/data/store.db on the mon pod's worker node for Rook. Look for old, rotated versions of *.log and MANIFEST*. Do NOT touch any *.sst files. Also check any other directories under /var/lib/rook and other directories on the same filesystem, often /var/log and /var/tmp are culprits. Your monitor hosts are; {{- range query "ceph_mon_metadata"}} - {{ .Labels.hostname }} {{- end }} - alert: CephMonClockSkew expr: ceph_health_detail{name="MON_CLOCK_SKEW"} == 1 for: 1m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#mon-clock-skew summary: Clock skew detected among monitors description: | Ceph monitors rely on closely synchronized time to maintain quorum and cluster consistency. This event indicates that time on at least one mon has drifted too far from the lead mon. Review cluster status with ceph -s. This will show which monitors are affected. Check the time sync status on each monitor host with "ceph time-sync-status" and the state and peers of your ntpd or chrony daemon. - name: osd rules: - alert: CephOSDDownHigh expr: count(ceph_osd_up == 0) / count(ceph_osd_up) * 100 >= 10 labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.4.1 annotations: summary: More than 10% of OSDs are down description: | {{ $value | humanize }}% or {{ with query "count(ceph_osd_up == 0)" }}{{ . | first | value }}{{ end }} of {{ with query "count(ceph_osd_up)" }}{{ . | first | value }}{{ end }} OSDs are down (>= 10%). The following OSDs are down: {{- range query "(ceph_osd_up * on(ceph_daemon) group_left(hostname) ceph_osd_metadata) == 0" }} - {{ .Labels.ceph_daemon }} on {{ .Labels.hostname }} {{- end }} - alert: CephOSDHostDown expr: ceph_health_detail{name="OSD_HOST_DOWN"} == 1 for: 5m labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.4.8 annotations: summary: An OSD host is offline description: | The following OSDs are down: {{- range query "(ceph_osd_up * on(ceph_daemon) group_left(hostname) ceph_osd_metadata) == 0" }} - {{ .Labels.hostname }} : {{ .Labels.ceph_daemon }} {{- end }} - alert: CephOSDDown expr: ceph_health_detail{name="OSD_DOWN"} == 1 for: 5m labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.4.2 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#osd-down summary: An OSD has been marked down description: | {{ $num := query "count(ceph_osd_up == 0)" | first | value }}{{ $s := "" }}{{ if gt $num 1.0 }}{{ $s = "s" }}{{ end }}{{ $num }} OSD{{ $s }} down for over 5mins. The following OSD{{ $s }} {{ if eq $s "" }}is{{ else }}are{{ end }} down: {{- range query "(ceph_osd_up * on(ceph_daemon) group_left(hostname) ceph_osd_metadata) == 0"}} - {{ .Labels.ceph_daemon }} on {{ .Labels.hostname }} {{- end }} - alert: CephOSDNearFull expr: ceph_health_detail{name="OSD_NEARFULL"} == 1 for: 5m labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.4.3 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#osd-nearfull summary: OSD(s) running low on free space (NEARFULL) description: | One or more OSDs have reached the NEARFULL threshold Use 'ceph health detail' and 'ceph osd df' to identify the problem. To resolve, add capacity to the affected OSD's failure domain, restore down/out OSDs, or delete unwanted data. - alert: CephOSDFull expr: ceph_health_detail{name="OSD_FULL"} > 0 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.4.6 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#osd-full summary: OSD full, writes blocked description: | An OSD has reached the FULL threshold. Writes to pools that share the affected OSD will be blocked. Use 'ceph health detail' and 'ceph osd df' to identify the problem. To resolve, add capacity to the affected OSD's failure domain, restore down/out OSDs, or delete unwanted data. - alert: CephOSDBackfillFull expr: ceph_health_detail{name="OSD_BACKFILLFULL"} > 0 for: 1m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#osd-backfillfull summary: OSD(s) too full for backfill operations description: | An OSD has reached the BACKFILL FULL threshold. This will prevent rebalance operations from completing. Use 'ceph health detail' and 'ceph osd df' to identify the problem. To resolve, add capacity to the affected OSD's failure domain, restore down/out OSDs, or delete unwanted data. - alert: CephOSDTooManyRepairs expr: ceph_health_detail{name="OSD_TOO_MANY_REPAIRS"} == 1 for: 30s labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#osd-too-many-repairs summary: OSD reports a high number of read errors description: | Reads from an OSD have used a secondary PG to return data to the client, indicating a potential failing disk. - alert: CephOSDTimeoutsPublicNetwork expr: ceph_health_detail{name="OSD_SLOW_PING_TIME_FRONT"} == 1 for: 1m labels: severity: warning type: ceph_default annotations: summary: Network issues delaying OSD heartbeats (public network) description: | OSD heartbeats on the cluster's 'public' network (frontend) are running slow. Investigate the network for latency or loss issues. Use 'ceph health detail' to show the affected OSDs. - alert: CephOSDTimeoutsClusterNetwork expr: ceph_health_detail{name="OSD_SLOW_PING_TIME_BACK"} == 1 for: 1m labels: severity: warning type: ceph_default annotations: summary: Network issues delaying OSD heartbeats (cluster network) description: | OSD heartbeats on the cluster's 'cluster' network (backend) are running slow. Investigate the network for latency or loss issues. Use 'ceph health detail' to show the affected OSDs. - alert: CephOSDInternalDiskSizeMismatch expr: ceph_health_detail{name="BLUESTORE_DISK_SIZE_MISMATCH"} == 1 for: 1m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#bluestore-disk-size-mismatch summary: OSD size inconsistency error description: | One or more OSDs have an internal inconsistency between metadata and the size of the device. This could lead to the OSD(s) crashing in future. You should redeploy the affected OSDs. - alert: CephDeviceFailurePredicted expr: ceph_health_detail{name="DEVICE_HEALTH"} == 1 for: 1m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#id2 summary: Device(s) predicted to fail soon description: | The device health module has determined that one or more devices will fail soon. To review device status use 'ceph device ls'. To show a specific device use 'ceph device info '. Mark the OSD out so that data may migrate to other OSDs. Once the OSD has drained, destroy the OSD, replace the device, and redeploy the OSD. - alert: CephDeviceFailurePredictionTooHigh expr: ceph_health_detail{name="DEVICE_HEALTH_TOOMANY"} == 1 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.4.7 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#device-health-toomany summary: Too many devices are predicted to fail, unable to resolve description: | The device health module has determined that devices predicted to fail can not be remediated automatically, since too many OSDs would be removed from the cluster to ensure performance and availabililty. Prevent data integrity issues by adding new OSDs so that data may be relocated. - alert: CephDeviceFailureRelocationIncomplete expr: ceph_health_detail{name="DEVICE_HEALTH_IN_USE"} == 1 for: 1m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#device-health-in-use summary: Device failure is predicted, but unable to relocate data description: | The device health module has determined that one or more devices will fail soon, but the normal process of relocating the data on the device to other OSDs in the cluster is blocked. Ensure that the cluster has available free space. It may be necessary to add capacity to the cluster to allow the data from the failing device to successfully migrate, or to enable the balancer. - alert: CephOSDFlapping expr: | ( rate(ceph_osd_up[5m]) * on(ceph_daemon) group_left(hostname) ceph_osd_metadata ) * 60 > 1 labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.4.4 annotations: documentation: https://docs.ceph.com/en/latest/rados/troubleshooting/troubleshooting-osd#flapping-osds summary: Network issues are causing OSDs to flap (mark each other down) description: > OSD {{ $labels.ceph_daemon }} on {{ $labels.hostname }} was marked down and back up {{ $value | humanize }} times once a minute for 5 minutes. This may indicate a network issue (latency, packet loss, MTU mismatch) on the cluster network, or the public network if no cluster network is deployed. Check network stats on the listed host(s). - alert: CephOSDReadErrors expr: ceph_health_detail{name="BLUESTORE_SPURIOUS_READ_ERRORS"} == 1 for: 30s labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#bluestore-spurious-read-errors summary: Device read errors detected description: > An OSD has encountered read errors, but the OSD has recovered by retrying the reads. This may indicate an issue with hardware or the kernel. # alert on high deviation from average PG count - alert: CephPGImbalance expr: | abs( ( (ceph_osd_numpg > 0) - on (job) group_left avg(ceph_osd_numpg > 0) by (job) ) / on (job) group_left avg(ceph_osd_numpg > 0) by (job) ) * on (ceph_daemon) group_left(hostname) ceph_osd_metadata > 0.30 for: 5m labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.4.5 annotations: summary: PGs are not balanced across OSDs description: > OSD {{ $labels.ceph_daemon }} on {{ $labels.hostname }} deviates by more than 30% from average PG count. # alert on high commit latency...but how high is too high - name: mds rules: - alert: CephFilesystemDamaged expr: ceph_health_detail{name="MDS_DAMAGE"} > 0 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.5.1 annotations: documentation: https://docs.ceph.com/en/latest/cephfs/health-messages#cephfs-health-messages summary: CephFS filesystem is damaged. description: > Filesystem metadata has been corrupted. Data may be inaccessible. Analyze metrics from the MDS daemon admin socket, or escalate to support. - alert: CephFilesystemOffline expr: ceph_health_detail{name="MDS_ALL_DOWN"} > 0 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.5.3 annotations: documentation: https://docs.ceph.com/en/latest/cephfs/health-messages/#mds-all-down summary: CephFS filesystem is offline description: > All MDS ranks are unavailable. The MDS daemons managing metadata are down, rendering the filesystem offline. - alert: CephFilesystemDegraded expr: ceph_health_detail{name="FS_DEGRADED"} > 0 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.5.4 annotations: documentation: https://docs.ceph.com/en/latest/cephfs/health-messages/#fs-degraded summary: CephFS filesystem is degraded description: > One or more metadata daemons (MDS ranks) are failed or in a damaged state. At best the filesystem is partially available, at worst the filesystem is completely unusable. - alert: CephFilesystemMDSRanksLow expr: ceph_health_detail{name="MDS_UP_LESS_THAN_MAX"} > 0 for: 1m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/cephfs/health-messages/#mds-up-less-than-max summary: MDS daemon count is lower than configured description: > The filesystem's "max_mds" setting defines the number of MDS ranks in the filesystem. The current number of active MDS daemons is less than this value. - alert: CephFilesystemInsufficientStandby expr: ceph_health_detail{name="MDS_INSUFFICIENT_STANDBY"} > 0 for: 1m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/cephfs/health-messages/#mds-insufficient-standby summary: Ceph filesystem standby daemons too few description: > The minimum number of standby daemons required by standby_count_wanted is less than the current number of standby daemons. Adjust the standby count or increase the number of MDS daemons. - alert: CephFilesystemFailureNoStandby expr: ceph_health_detail{name="FS_WITH_FAILED_MDS"} > 0 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.5.5 annotations: documentation: https://docs.ceph.com/en/latest/cephfs/health-messages/#fs-with-failed-mds summary: MDS daemon failed, no further standby available description: > An MDS daemon has failed, leaving only one active rank and no available standby. Investigate the cause of the failure or add a standby MDS. - alert: CephFilesystemReadOnly expr: ceph_health_detail{name="MDS_HEALTH_READ_ONLY"} > 0 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.5.2 annotations: documentation: https://docs.ceph.com/en/latest/cephfs/health-messages#cephfs-health-messages summary: CephFS filesystem in read only mode due to write error(s) description: > The filesystem has switched to READ ONLY due to an unexpected error when writing to the metadata pool. Analyze the output from the MDS daemon admin socket, or escalate to support. - name: mgr rules: - alert: CephMgrModuleCrash expr: ceph_health_detail{name="RECENT_MGR_MODULE_CRASH"} == 1 for: 5m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.6.1 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#recent-mgr-module-crash summary: A manager module has recently crashed description: > One or more mgr modules have crashed and have yet to be acknowledged by an administrator. A crashed module may impact functionality within the cluster. Use the 'ceph crash' command to determine which module has failed, and archive it to acknowledge the failure. - alert: CephMgrPrometheusModuleInactive expr: up{job="ceph"} == 0 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.6.2 annotations: summary: The mgr/prometheus module is not available description: > The mgr/prometheus module at {{ $labels.instance }} is unreachable. This could mean that the module has been disabled or the mgr daemon itself is down. Without the mgr/prometheus module metrics and alerts will no longer function. Open a shell to an admin node or toolbox pod and use 'ceph -s' to to determine whether the mgr is active. If the mgr is not active, restart it, otherwise you can determine the mgr/prometheus module status with 'ceph mgr module ls'. If it is not listed as enabled, enable it with 'ceph mgr module enable prometheus'. - name: pgs rules: - alert: CephPGsInactive expr: ceph_pool_metadata * on(pool_id,instance) group_left() (ceph_pg_total - ceph_pg_active) > 0 for: 5m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.7.1 annotations: summary: One or more placement groups are inactive description: > {{ $value }} PGs have been inactive for more than 5 minutes in pool {{ $labels.name }}. Inactive placement groups are not able to serve read/write requests. - alert: CephPGsUnclean expr: ceph_pool_metadata * on(pool_id,instance) group_left() (ceph_pg_total - ceph_pg_clean) > 0 for: 15m labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.7.2 annotations: summary: One or more placement groups are marked unclean description: > {{ $value }} PGs have been unclean for more than 15 minutes in pool {{ $labels.name }}. Unclean PGs have not recovered from a previous failure. - alert: CephPGsDamaged expr: ceph_health_detail{name=~"PG_DAMAGED|OSD_SCRUB_ERRORS"} == 1 for: 5m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.7.4 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#pg-damaged summary: Placement group damaged; manual intervention needed description: > Scrubs have flagged at least one PG as damaged or inconsistent. Check to see which PG is affected, and attempt a manual repair if necessary. To list problematic placement groups, use 'ceph health detail' or 'rados list-inconsistent-pg '. To repair PGs use the 'ceph pg repair ' command. - alert: CephPGRecoveryAtRisk expr: ceph_health_detail{name="PG_RECOVERY_FULL"} == 1 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.7.5 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#pg-recovery-full summary: OSDs are too full for recovery description: > Data redundancy is at risk since one or more OSDs are at or above the 'full' threshold. Add capacity to the cluster, restore down/out OSDs, or delete unwanted data. - alert: CephPGUnavailableBlockingIO # PG_AVAILABILITY, but an OSD is not in a DOWN state expr: ((ceph_health_detail{name="PG_AVAILABILITY"} == 1) - scalar(ceph_health_detail{name="OSD_DOWN"})) == 1 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.7.3 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#pg-availability summary: PG is unavailable, blocking I/O description: > Data availability is reduced, impacting the cluster's ability to service I/O. One or more placement groups (PGs) are in a state that blocks I/O. - alert: CephPGBackfillAtRisk expr: ceph_health_detail{name="PG_BACKFILL_FULL"} == 1 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.7.6 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#pg-backfill-full summary: Backfill operations are blocked due to lack of free space description: > Data redundancy may be at risk due to lack of free space within the cluster. One or more OSDs have breached their 'backfillfull' threshold. Add more capacity, or delete unwanted data. - alert: CephPGNotScrubbed expr: ceph_health_detail{name="PG_NOT_SCRUBBED"} == 1 for: 5m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#pg-not-scrubbed summary: Placement group(s) have not been scrubbed description: | One or more PGs have not been scrubbed recently. Scrubs check metadata integrity, protecting against bit-rot. They check that metadata is consistent across data replicas. When PGs miss their scrub interval, it may indicate that the scrub window is too small, or PGs were not in a 'clean' state during the scrub window. You can manually initiate a scrub with: ceph pg scrub - alert: CephPGsHighPerOSD expr: ceph_health_detail{name="TOO_MANY_PGS"} == 1 for: 1m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks/#too-many-pgs summary: Placement groups per OSD is too high description: | The number of placement groups per OSD is too high (exceeds the mon_max_pg_per_osd setting). Check that the pg_autoscaler has not been disabled for any pools with 'ceph osd pool autoscale-status', and that the profile selected is appropriate. You may also adjust the target_size_ratio of a pool to guide the autoscaler based on the expected relative size of the pool ('ceph osd pool set cephfs.cephfs.meta target_size_ratio .1') or set the pg_autoscaler mode to "warn" and adjust pg_num appropriately for one or more pools. - alert: CephPGNotDeepScrubbed expr: ceph_health_detail{name="PG_NOT_DEEP_SCRUBBED"} == 1 for: 5m labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#pg-not-deep-scrubbed summary: Placement group(s) have not been deep scrubbed description: | One or more PGs have not been deep scrubbed recently. Deep scrubs protect against bit-rot. They compare data replicas to ensure consistency. When PGs miss their deep scrub interval, it may indicate that the window is too small or PGs were not in a 'clean' state during the deep-scrub window. You can manually initiate a deep scrub with: ceph pg deep-scrub - name: nodes rules: - alert: CephNodeRootFilesystemFull expr: node_filesystem_avail_bytes{mountpoint="/"} / node_filesystem_size_bytes{mountpoint="/"} * 100 < 5 for: 5m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.8.1 annotations: summary: Root filesystem is dangerously full description: > Root volume is dangerously full: {{ $value | humanize }}% free. # alert on packet errors and drop rate - alert: CephNodeNetworkPacketDrops expr: | ( increase(node_network_receive_drop_total{device!="lo"}[1m]) + increase(node_network_transmit_drop_total{device!="lo"}[1m]) ) / ( increase(node_network_receive_packets_total{device!="lo"}[1m]) + increase(node_network_transmit_packets_total{device!="lo"}[1m]) ) >= 0.0001 or ( increase(node_network_receive_drop_total{device!="lo"}[1m]) + increase(node_network_transmit_drop_total{device!="lo"}[1m]) ) >= 10 labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.8.2 annotations: summary: One or more NICs reports packet drops description: > Node {{ $labels.instance }} experiences packet drop > 0.01% or > 10 packets/s on interface {{ $labels.device }}. - alert: CephNodeNetworkPacketErrors expr: | ( increase(node_network_receive_errs_total{device!="lo"}[1m]) + increase(node_network_transmit_errs_total{device!="lo"}[1m]) ) / ( increase(node_network_receive_packets_total{device!="lo"}[1m]) + increase(node_network_transmit_packets_total{device!="lo"}[1m]) ) >= 0.0001 or ( increase(node_network_receive_errs_total{device!="lo"}[1m]) + increase(node_network_transmit_errs_total{device!="lo"}[1m]) ) >= 10 labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.8.3 annotations: summary: One or more NICs reports packet errors description: > Node {{ $labels.instance }} experiences packet errors > 0.01% or > 10 packets/s on interface {{ $labels.device }}. # Restrict to device names beginning with '/' to skip false alarms from # tmpfs, overlay type filesystems - alert: CephNodeDiskspaceWarning expr: | predict_linear(node_filesystem_free_bytes{device=~"/.*"}[2d], 3600 * 24 * 5) * on(instance) group_left(nodename) node_uname_info < 0 labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.8.4 annotations: summary: Host filesystem free space is low description: > Mountpoint {{ $labels.mountpoint }} on {{ $labels.nodename }} will be full in less than 5 days based on the 48 hour trailing fill rate. - alert: CephNodeInconsistentMTU expr: node_network_mtu_bytes{device!="lo"} * (node_network_up{device!="lo"} > 0) != on() group_left() (quantile(0.5, node_network_mtu_bytes{device!="lo"})) labels: severity: warning type: ceph_default annotations: summary: MTU settings across hosts are inconsistent description: > Node {{ $labels.instance }} has a different MTU size ({{ $value }}) than the median value on device {{ $labels.device }}. - name: pools rules: - alert: CephPoolGrowthWarning expr: | (predict_linear((max(ceph_pool_percent_used) without (pod, instance))[2d:1h], 3600 * 24 * 5) * on(pool_id) group_right ceph_pool_metadata) >= 95 labels: severity: warning type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.9.2 annotations: summary: Pool growth rate may soon exceed capacity description: > Pool '{{ $labels.name }}' will be full in less than 5 days assuming the average fill-up rate of the past 48 hours. - alert: CephPoolBackfillFull expr: ceph_health_detail{name="POOL_BACKFILLFULL"} > 0 labels: severity: warning type: ceph_default annotations: summary: Free space in a pool is too low for recovery/backfill description: > A pool is approaching the near full threshold, which will prevent recovery/backfill from completing. Consider adding more capacity. - alert: CephPoolFull expr: ceph_health_detail{name="POOL_FULL"} > 0 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.9.1 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#pool-full summary: Pool is full - writes are blocked description: | A pool has reached its MAX quota, or OSDs supporting the pool have reached the FULL threshold. Until this is resolved, writes to the pool will be blocked. Pool Breakdown (top 5) {{- range query "topk(5, sort_desc(ceph_pool_percent_used * on(pool_id) group_right ceph_pool_metadata))" }} - {{ .Labels.name }} at {{ .Value }}% {{- end }} Increase the pool's quota, or add capacity to the cluster then increase the pool's quota (e.g. ceph osd pool set quota max_bytes ) - alert: CephPoolNearFull expr: ceph_health_detail{name="POOL_NEAR_FULL"} > 0 for: 5m labels: severity: warning type: ceph_default annotations: summary: One or more Ceph pools are nearly full description: | A pool has exceeded the warning (percent full) threshold, or OSDs supporting the pool have reached the NEARFULL threshold. Writes may continue, but you are at risk of the pool going read-only if more capacity isn't made available. Determine the affected pool with 'ceph df detail', looking at QUOTA BYTES and STORED. Increase the pool's quota, or add capacity to the cluster then increase the pool's quota (e.g. ceph osd pool set quota max_bytes ). Also ensure that the balancer is active. - name: healthchecks rules: - alert: CephSlowOps expr: ceph_healthcheck_slow_ops > 0 for: 30s labels: severity: warning type: ceph_default annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#slow-ops summary: OSD operations are slow to complete description: > {{ $value }} OSD requests are taking too long to process (osd_op_complaint_time exceeded) # Object related events - name: rados rules: - alert: CephObjectMissing expr: (ceph_health_detail{name="OBJECT_UNFOUND"} == 1) * on() (count(ceph_osd_up == 1) == bool count(ceph_osd_metadata)) == 1 for: 30s labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.10.1 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks#object-unfound summary: Object(s) marked UNFOUND description: | The latest version of a RADOS object can not be found, even though all OSDs are up. I/O requests for this object from clients will block (hang). Resolving this issue may require the object to be rolled back to a prior version manually, and manually verified. # Generic - name: generic rules: - alert: CephDaemonCrash expr: ceph_health_detail{name="RECENT_CRASH"} == 1 for: 1m labels: severity: critical type: ceph_default oid: 1.3.6.1.4.1.50495.1.2.1.1.2 annotations: documentation: https://docs.ceph.com/en/latest/rados/operations/health-checks/#recent-crash summary: One or more Ceph daemons have crashed, and are pending acknowledgement description: | One or more daemons have crashed recently, and need to be acknowledged. This notification ensures that software crashes do not go unseen. To acknowledge a crash, use the 'ceph crash archive ' command.