John Spray

ARM tests are okay now - green run from last night here https://buildkite.com/redpanda/vtools/builds/6732#0186e6ee-35c5-4ca7-a43b-6a2c8eb474ce

Reopening because it's not clear if #7297 is the end of the story or if there is more work here to deal with how the consensus object got into the...

I tested this by hand on ARM to check if it was some environment issue like slow disks. It looks like a genuine test failure. The leader balancer is waking...

(removed myself as reviewer, assuming that was an automated request. Feel free to re-add me if you need my :eyes: )

> Why is it necessary to intermingle the cancellation changes with the more efficient timeline lookups? The cancellation changes are to go from respecting one timeline's cancellation to respecting multiple:...

>I wonder if the "keep gate open" aspect could have been solved more cleanly by another level of indirection. > What we're really keeping open is our PageServerHandler instance's view...

I've cleaned this up to cover what we did in Q4 -- will publish a smaller follow-on RFC that describes shard splitting (Q1).

Two things before merging: - Please check in with Heikki: he was a reviewer on the previous PR - Please write a proper description, as this will end up in...

> How to test this PR? There's a circular dependency: this PR depends on the changes in the rest of the stack for testing (https://github.com/neondatabase/neon/pull/6380), and they depend on this...

Strange, I can see conversations when I scroll up (sometimes github collapses part of the history and you have to click a divider to expand it):