You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Use case. Why is this important?
I have a use case from my experience. I had huge spikes on checkpoints. At first, I thought the checkpointer was late and that I needed to increase max_wal_size/checkpoint_timeout, but it didn't help. I accidentally noticed that wal_fpi increased along with the spikes. To prove this, my workaround was to sample it with an SQL exporter, and it turned out to be true. The core problem was that we had a table with a GUID index and a huge insert workload, which rapidly dirties almost all the index's pages, triggering massive wal_fpi on checkpoints. Having at least wal_fpi would be very useful to quickly discover such problems. Also, I think it would be useful to have the entire pg_stat_wal in the exporter, but for other columns, I don't have such motivation.
The text was updated successfully, but these errors were encountered:
Add support for
pg_stat_wal
Use case. Why is this important?
I have a use case from my experience. I had huge spikes on checkpoints. At first, I thought the checkpointer was late and that I needed to increase max_wal_size/checkpoint_timeout, but it didn't help. I accidentally noticed that wal_fpi increased along with the spikes. To prove this, my workaround was to sample it with an SQL exporter, and it turned out to be true. The core problem was that we had a table with a GUID index and a huge insert workload, which rapidly dirties almost all the index's pages, triggering massive wal_fpi on checkpoints. Having at least wal_fpi would be very useful to quickly discover such problems. Also, I think it would be useful to have the entire pg_stat_wal in the exporter, but for other columns, I don't have such motivation.
The text was updated successfully, but these errors were encountered: