[xCluster]: Change xCluster Metrics Collection Interval from 1s to 15s #9417
Comments
|
Taking this up. |
|
talked to @sanketkedia offline, since he's short term focusing on PITR for the next 2.6, we can take this on in the short term xcluster work as well |
rahuldesirazu
added a commit
that referenced
this issue
Aug 20, 2021
… and bump update_metrics_interval_ms to 15000 Summary: #9662: Currently a user has to set enable_log_retention_by_op_idx to true for xCluster setups. We can default this flag to true, since we only change WAL retention policy for entries in the cdc_state table. So for setups with no xCluster enabled, changing this flag will have no effect. #9417: We currently set --update_metrics_interval_ms = 1000, which means we collect metrics every 1s. This operation opens the cdc_state table and loops through all tablets under-replication, which can be an expensive operation. Since we don't need such fine-grained metrics, we can change this to 15000 to collect every 15s. Test Plan: Jenkins Reviewers: nicolas, jhe Reviewed By: jhe Subscribers: ybase, bogdan Differential Revision: https://phabricator.dev.yugabyte.com/D12638
rahuldesirazu
added a commit
that referenced
this issue
Aug 31, 2021
… true by default and bump update_metrics_interval_ms to 15000 Summary: #9662: Currently a user has to set enable_log_retention_by_op_idx to true for xCluster setups. We can default this flag to true, since we only change WAL retention policy for entries in the cdc_state table. So for setups with no xCluster enabled, changing this flag will have no effect. #9417: We currently set --update_metrics_interval_ms = 1000, which means we collect metrics every 1s. This operation opens the cdc_state table and loops through all tablets under-replication, which can be an expensive operation. Since we don't need such fine-grained metrics, we can change this to 15000 to collect every 15s. Original Commit : D12638 /2933bf506fbf7b6ac8fe6d6715fa4c03b8dcd550 Test Plan: Jenkins: rebase: 2.6 Reviewers: nicolas, jhe Reviewed By: jhe Subscribers: ybase, bogdan Differential Revision: https://phabricator.dev.yugabyte.com/D12804
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
We currently set
--update_metrics_interval_ms = 1000, which means we collect metrics every 1s. This operation opens the cdc_state table and loops through all tablets under-replication, which can be an expensive operation. Since we don't need such fine-grained metrics, we can change this to 15000 to collect every 15s.The text was updated successfully, but these errors were encountered: