Merge "Add alerting rules for RGW multisite deployments"
This commit is contained in:
commit
6ae78a6e6c
|
@ -633,3 +633,50 @@ groups:
|
|||
oid: "1.3.6.1.4.1.50495.1.2.1.1.2"
|
||||
severity: "critical"
|
||||
type: "ceph_default"
|
||||
- name: "rgwmultisite"
|
||||
rules:
|
||||
- alert: "CephRGWMultisiteFetchError"
|
||||
annotations:
|
||||
description: "Unsuccessful Object Replications from source zone threshold has been exceeded. The threshold is defined as 2 errors per 15min"
|
||||
summary: "Unsuccessful Object Replications from Source Zone Threshold Exceeded"
|
||||
expr: "increase(ceph_data_sync_from_zone_fetch_errors[15m]) > 2"
|
||||
for: "5m"
|
||||
labels:
|
||||
severity: "warning"
|
||||
type: "ceph_default"
|
||||
- alert: "CephRGWMultisitePollError"
|
||||
annotations:
|
||||
description: "Unsuccessful Object Replications from Source Zone Threshold Exceeded. The threshold is defined as 2 errors per 15min"
|
||||
summary: "Unsuccessful Object Replications from Source Zone Threshold Exceeded"
|
||||
expr: "increase(ceph_data_sync_from_zone_poll_errors[15m]) > 2"
|
||||
for: "5m"
|
||||
labels:
|
||||
severity: "warning"
|
||||
type: "ceph_default"
|
||||
- alert: "CephRGWMultisiteFetchErrorCritical"
|
||||
annotations:
|
||||
description: "Critical: Unsuccessful Object Replications from source zone threshold has been exceeded. The threshold is defined as 50 errors per 15min"
|
||||
summary: "Critical: Unsuccessful Object Replications from Source Zone Threshold Exceeded"
|
||||
expr: "increase(ceph_data_sync_from_zone_fetch_errors[15m]) > 50"
|
||||
for: "5m"
|
||||
labels:
|
||||
severity: "critical"
|
||||
type: "ceph_default"
|
||||
- alert: "CephRGWMultisitePollErrorCritical"
|
||||
annotations:
|
||||
description: "Critical: Unsuccessful Object Replications from source zone threshold has been exceeded. The threshold is defined as 50 errors per 15min"
|
||||
summary: "Critical: Unsuccessful Object Replications from Source Zone Threshold Exceeded"
|
||||
expr: "increase(ceph_data_sync_from_zone_poll_errors[15m]) > 50"
|
||||
for: "5m"
|
||||
labels:
|
||||
severity: "critical"
|
||||
type: "ceph_default"
|
||||
- alert: "CephRGWMultisitePollLatency"
|
||||
annotations:
|
||||
description: "Latency for poll request threshold exceeded. The threshold is defined as 600s latency per 15min"
|
||||
summary: "Poll Request Latency Threshold Exceeded"
|
||||
expr: "increase(ceph_data_sync_from_zone_poll_latency_sum[15m]) > 600"
|
||||
for: "5m"
|
||||
labels:
|
||||
severity: "warning"
|
||||
type: "ceph_default"
|
||||
|
|
Loading…
Reference in New Issue