Chris Bandy

Instead of that sentence, I want to recommend (something like) the following: > The following actions are supported: > - `pause` - pause when recovery target is reached on a...

I discovered another wrinkle while researching this topic. Before PostgreSQL 12, the "pause-means-shutdown" behavior requires an explicit `recovery_target_action=pause` in `recovery.conf`. Notice `recoveryTargetActionSet` below: https://github.com/postgres/postgres/blob/REL_11_7/src/backend/access/transam/xlog.c#L5551-L5554 ```c if (recoveryTargetAction == RECOVERY_TARGET_ACTION_PAUSE &&...

The PostgreSQL releases since 12 have behaved consistently, so I think the following covers all the issues: 1. The `promote` and `shutdown` values work correctly regardless of `hot_standby`. 2. The...

Last I checked, those variables affect how the container is built: https://github.com/zalando/spilo/blob/7516aff6565b41b9f38840f696a563e8f87c34ad/postgres-appliance/Dockerfile#L164-L185 To be clear, are you *building* the container (`docker build`) with those variables and/or *running* (`docker run`) with...

Looks like this can be closed now that #130 is merged?

Both Lambda and Aurora Serverless sound like cases where connections should be handled carefully. In the Sequel issue, I see you are using a connection pool. Questions that come to...

> However, I wasn't able to get much insight into the Aurora cluster and what was happening from that point of view as the connection hung. You'll need to go...

Hmm. All of the traces appear cutoff. Perhaps there needs to be an `f.flush` or `f.fsync` at the end of the block (before the file is closed.) Each of the...

@aniketschneider This sounds a little bit like it stalls after the same amount of bytes transferred. What happens if you try a query like the following? It returns 1GiB of...

When a particular batch size hangs, does it consistently/repeatedly do so on a particular `slice_num`?