Reduce max_replication_write_lag to 10Mb

[knizhnik]

No description provided.

[petuhovskiy]

We probably should also update it in the console, I found several references there:

• https://github.com/neondatabase/cloud/blob/3942abd11f046e175dce3a977970671033ac184e/goapp/internal/controllers/publicapiv2/settings.go
• https://github.com/neondatabase/cloud/blob/a6039b88da608c36e9cf8334e015229018f5a834/goapp/internal/model/project.go

[lubennikovaav]

@knizhnik , I added small fix to tests and pushed to trigger tests again.

[hlinnaka]

Rebased

[stepashka]

@knizhnik , any next steps planned for this?
have we merged this?
was it causing any troubles?

[stepashka]

we've increased the lag to 15 MBs, maybe we can close this PR?

[knizhnik]

we've increased the lag to 15 MBs, maybe we can close this PR?

We have not changed default value.
There is really IMHO stupid situation that some our default settings, particularly wal_log_hints and max_replication write_lag, which are different from what are actually used for tests and in staging/production.

[MMeent]

So, if I understand correctly, we currently use values <100MB effectively everywhere, and only this default is exceptionally large. If that understanding is correct, I think this is OK to merge.

[hlinnaka]

Are these values now the same that we use in production?

[MMeent]

This new default is off by 5MB from our production settings for PG15:

$ SHOW max_replication_lag;
 max_replication_write_lag 
---------------------------
 15MB