David Steele

One thing you might consider is increasing the WAL segment size. This can be very beneficial for performance. You can do this with pg_resetwal, but it requires an outage and...

We've looked into doing something like this before but we ran into some blockers. I can't remember what they were, though. I think @Tostino was woking on it?

This looks related to #987, @cbandy. Honestly we just haven't gotten around to updating the docs. If someone could test the configurations and submit a PR with updates to the...

We should test this before making a code change. So, do a backup with `archive-copy`, do the restore, remove `recovery.signal` and make sure it comes up as a primary with...

> I tested it on my end and Postgres was able to come up as a primary without needing to perform a timeline switch Don't ever do this. You run...

> here's a proposal to integrate an annotation system. This looks like a very useful feature! > Hm, I'll look at the src/info/manifest.c conflict Yeah, you picked a bad time...

> Those annotations (created by the annotate command) could be lost if backup.info is lost/broken for any reason... I think that'll be the least of their worries since the repo...

I think we can update annotations by writing a `backup.annotation` file into the backup directory. Then look for that file when reconstructing `backup.info`. Let's no worry yet about how this...

Thinking some more -- the problem with writing a `backup.annotation` file is that it would require a new format and need an object to support it. I'm actively working on...

> So I tend to only display it with info --set=.. Hmmm. It seems to me that annotations are meant to be used in an automated fashion and text does...