Excuse my ignorance but is there an easy way to monitor the thread commit times as this is critical for our app ? For example when a thread is in COMMIT state for more than X seconds. We are running Percona XtraDB Cluster 5.6.51-91.0-56-log.
Can I use mysql_info_schema_write_query_response_time_seconds_bucket for that purpose given that the query response time plugin is enabled ?
Having said that, are you aware of what’s the impact of the query response time plugin on the cluster ? I have 5 nodes cluster with two active nodes and the plugin works just fine on non-prod but I’m a bit hesitant to enable it on prod.
We are running Percona XtraDB Cluster 5.6.51-91.0-56-log.
Note that 5.6 and 5.7 have already reached EOL (end of life), and the query response time plugin has not been ported to 8.0. PMM does offer a dashboard to view data gathered with this plugin, though:
MySQL Query Response Time Details.
Having said that, are you aware of what’s the impact of the query response time plugin on the cluster ? I have 5 nodes cluster with two active nodes and the plugin works just fine on non-prod but I’m a bit hesitant to enable it on prod.
I suggest you to test this in a similar environment, because it will depend on your workload and hardware specs. Having said this, you should have COMMIT information already, so I would first check if that is enough.
I can’t find an easier way to do that and decided to use the active threads instead. So that when there is an issue or flow control the number of active threads will start rising (we have a predictable number if active connections) will raise an alert. Thanks for your time.
