-
Bug
-
Resolution: Done
-
Undefined
-
None
-
False
-
False
-
NEW
-
VERIFIED
-
-
-
Logging (LogExp) - Sprint 213
There are two dashboards for Logging.
1. Logging / Elasticsearch – which presents the number of Elasticsearch cluster Primary Shards and All Shards.
max(es_cluster_shards_number{cluster="elasticsearch",type="active"})
2. OpenShift Logging – which presents a number of shards. But this value is 3x bigger than reality. It is counted by:
sum(sum by(instance)(es_cluster_shards_number{type='active_primary'}))
and the result in my case is 33 shards while the reality is: 11 shards (+/-). So when we check the number of shards in the OpenShift Logging with:
$ oc exec $ES_POD -- shards or $ ec exec $ES_PODS -- es_util --query=_cat/shards
We can see that the number of shards in the ES cluster is less. Looks like the metric is printing all primary shards for each ES pod and processing sum on it while the output is always for the whole ES cluster.
- clones
-
LOG-2048 Dashboard for OpenShift Logging in WebConsole shows incorrect number of shards
- Closed
- links to