Lawrence Jones

Another example in #44

This actually addresses #601, though not via any of the remediations recommended in that issue.

Hey @rearden-steel @frank111, I'm not sure how this is related to the wal keep segments GUC? This change is about recovering from forked timelines rather than having insufficient wal to...

Hey @rearden-steel, I'm.midway through making this work. It needs integration tests which are what I've been working on, and to provide a configurable variable that enables/disables checkpointing. I'm out at...

From where I got to last time, it's difficult to simulate a Postgres that takes a while to checkpoint after a promotion. While pg_ctl promote won't force a synchronous checkpoint...

@sgotti @rearden-steel this is ready for review now. Instead of adding this logic into an existing test, I've created another test with just two keepers that verifies we're safe from...

Hey @sgotti! I've been rubbish here: PRs were left hanging while I went on holiday 🏖 I'm going to reply to each bit of your review separately. Hopefully we can...

@sgotti I know you're probably busy on a lot of other things right now, but would really appreciate your feedback for this change. It matters quite a lot to us...

Hey @sgotti! > So I don't want to stop this PR trying to find a cleanliness that's not really possible. Lets go ahead with your solution! (in the end we...

Hey @viggy28, if you want to take this over then please let me know. I’m not working on stolon things right now so if you need this, I’ll be happy...