use correct units for MDSCacheUsageHigh alerts

ceph-mds-mem-rss, used in MDSCacheUsageHigh alert, is in KB rather than bytes. This PR converts the ceph-mds-mem-rss metric value to bytes before comparing it with the mds pod memory limit. Signed-off-by: sp98 <[email protected]>
red-hat-storage · Jan 4, 2024 · 27a90ff · 27a90ff
1 parent e4fb5bd
commit 27a90ff
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/metrics/deploy/prometheus-ocs-rules.yaml b/metrics/deploy/prometheus-ocs-rules.yaml
@@ -282,7 +282,7 @@ spec:
         message: High MDS cache usage for the daemon {{ $labels.ceph_daemon }}.
         severity_level: error
       expr: |
-        ceph_mds_mem_rss / on(ceph_daemon) group_left(job)(label_replace(kube_pod_container_resource_requests{container="mds", resource="memory"}, "ceph_daemon", "mds.$1", "pod", "rook-ceph-mds-(.*)-(.*)") * .5) > .95
+        (ceph_mds_mem_rss * 1000) / on(ceph_daemon) group_left(job)(label_replace(kube_pod_container_resource_requests{container="mds", resource="memory"}, "ceph_daemon", "mds.$1", "pod", "rook-ceph-mds-(.*)-(.*)") * .5) > .95
       for: 5m
       labels:
         severity: critical

diff --git a/metrics/mixin/alerts/perf.libsonnet b/metrics/mixin/alerts/perf.libsonnet
@@ -7,7 +7,7 @@
           {
             alert: 'MDSCacheUsageHigh',
             expr: |||
-              ceph_mds_mem_rss / on(ceph_daemon) group_left(job)(label_replace(kube_pod_container_resource_requests{container="mds", resource="memory"}, "ceph_daemon", "mds.$1", "pod", "rook-ceph-mds-(.*)-(.*)") * .5) > .95
+              (ceph_mds_mem_rss * 1000) / on(ceph_daemon) group_left(job)(label_replace(kube_pod_container_resource_requests{container="mds", resource="memory"}, "ceph_daemon", "mds.$1", "pod", "rook-ceph-mds-(.*)-(.*)") * .5) > .95
             ||| % $._config,
             'for': $._config.mdsCacheUsageAlertTime,
             labels: {