John Spray

Do we like this as a cause for https://github.com/redpanda-data/redpanda/issues/7343 ? It seems odd that the issue would manifest on upgrade if this is a long-standing bug in Redpanda.

> I think we should make changes to return type in a different PR to reduce the change surface area but happy to take a stab at it now if...

> If we do run into that case because of truncation then the segment collector on next run fast forwards the begin offset to the start of the manifest and...

The backported test is failing (i.e. presumably this is when it tries to upgrade from 22.1.x to 22.2.x) Given that most of these upgrades in the field have already happened,...

Ah-ha... but we could still fix this by changing the partitioner, right? If we used a partitioner than insisted on using deterministic partition choice based on key, it would wait...

@ztlpn is there a particular PR this one is waiting for? I'm not up to speed on the controller change you mentioned.

@andijcr please can you work on this when https://github.com/redpanda-data/redpanda/issues/6843 is done

Thinking about delivering this, I think there are probably ~3 PRs: 1. In the places where we aim to test a particular version hop (e.g. before/after a feature is introduced),...

@joejulian @dotnwat is this ticket still relvant?

@andrwng this may be of interest to you when looking at local storage trimming on disk pressure. It cleans up the structure a bit so that we have local_monitor running...