monitoring: Fix pool capacity incorrect

Signed-off-by: James Cheng <james59988@gmail.com>
(cherry picked from commit 1b980ef88c5f8f45b5b39b79c395a66233908266)

diff --git a/monitoring/grafana/dashboards/pool-detail.json b/monitoring/grafana/dashboards/pool-detail.json
index 14e468a5274edf2a8cc873cbd7797b26d5eb63a4..f7e90da7e28d83a7b5042f13597cfe7a50c8f180 100644 (file)
--- a/monitoring/grafana/dashboards/pool-detail.json
+++ b/monitoring/grafana/dashboards/pool-detail.json
@@ -101,7 +101,7 @@
       "tableColumn": "",
       "targets": [
         {
-          "expr": "(ceph_pool_stored / ceph_pool_max_avail) * on(pool_id) group_left(instance,name) ceph_pool_metadata{name=~\"$pool_name\"}",
+          "expr": "(ceph_pool_stored / (ceph_pool_stored + ceph_pool_max_avail)) * on(pool_id) group_left(instance,name) ceph_pool_metadata{name=~\"$pool_name\"}",
           "format": "time_series",
           "intervalFactor": 1,
           "refId": "A"
@@ -182,7 +182,7 @@
       "tableColumn": "",
       "targets": [
         {
-          "expr": "((ceph_pool_max_avail - ceph_pool_stored) / deriv(ceph_pool_stored[6h])) * on(pool_id) group_left(instance,name) ceph_pool_metadata{name=~\"$pool_name\"} > 0",
+          "expr": "(ceph_pool_max_avail / deriv(ceph_pool_stored[6h])) * on(pool_id) group_left(instance,name) ceph_pool_metadata{name=~\"$pool_name\"} > 0",
           "format": "time_series",
           "intervalFactor": 1,
           "refId": "A"

diff --git a/monitoring/prometheus/alerts/ceph_default_alerts.yml b/monitoring/prometheus/alerts/ceph_default_alerts.yml
index 4c37d7f9c9b0a949296c772263edd7b9949509fc..5eb31e909874f4a4240d42fbdc2580447e4fe23b 100644 (file)
--- a/monitoring/prometheus/alerts/ceph_default_alerts.yml
+++ b/monitoring/prometheus/alerts/ceph_default_alerts.yml
@@ -203,7 +203,7 @@ groups:
     rules:
       - alert: pool full
         expr: |
-          ceph_pool_stored / ceph_pool_max_avail
+          ceph_pool_stored / (ceph_pool_stored + ceph_pool_max_avail)
           * on(pool_id) group_right ceph_pool_metadata * 100 > 90
         labels:
           severity: critical