Chi-Hsuan Huang created HDDS-16719:
--------------------------------------
Summary: Use MutableGaugeLong for snapshot and tenant counts in
OMMetrics
Key: HDDS-16719
URL: https://issues.apache.org/jira/browse/HDDS-16719
Project: Apache Ozone
Issue Type: Bug
Components: Ozone Manager
Reporter: Chi-Hsuan Huang
Follow\-up to HDDS\-16611.
{{OMMetrics}} declares {{numSnapshotActive}}, {{numSnapshotDeleted}} and
{{numTenants}} as {{MutableCounterLong}}, but they behave like gauges:
\* {{decNumSnapshotActive\(\)}}, {{decNumSnapshotDeleted\(\)}} and
{{decNumTenants\(\)}} decrement with {{incr\(\-1\)}}.
\* {{setNumSnapshotActive\(\)}} and {{setNumSnapshotDeleted\(\)}} emulate a set
via {{incr\(num \- currVal\)}}. The caller in {{OzoneManager}} already names
these values {{activeGauge}} and {{deletedGauge}}.
As a result, {{/prom}} exports them as {{# TYPE counter}}, and Prometheus
treats any decrease as a counter reset. The bundled OM Snapshot dashboard only
plots the raw values, so it is not affected today. Any {{rate\(\)}} or
{{increase\(\)}} query on these metrics would report wrong values.
Proposed changes:
\* Switch the three fields to {{MutableGaugeLong}} and use {{set\(\)}} in the
two setters.
\* Update {{TestOmMetrics}} to read {{NumSnapshotActive}} and
{{NumSnapshotDeleted}} with {{getLongGauge}} instead of {{getLongCounter}}.
Metric names in {{/jmx}} and {{/prom}} stay the same.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
