John Spray

When merging `main`, I noticed that changes from #8648 were calling `is_valid_key_on_write_path` during ingest at some points where we can't do that any more (because they get converted down to...

Can you update existing tests to invoke the API via the storage controller instead of directly on the pageserver, so that we have some coverage of this?

- Calling sync before cleaning up ephemeral layers is somewhat wasteful - However, ephemeral layer cleanup happens relatively late in startup, after loading timeline metadata. - This should only be...

This is complete once https://github.com/neondatabase/neon/pull/8648 lands (with nice behavior for stopping ingest if it sees a key it can't read)

As of https://github.com/neondatabase/neon/pull/9383, we will warn on a suspicious old index. Next steps: before end of next week, check logs for these warnings. There shouldn't be any. Then write another...

So there are no `Found a newer index while loading an old one` warnings, which is good and as expected. However, the `Loaded an old index, checking for other indices...`...

@VladLazar your call how much we should evolve this RFC: it's fairly high level and I think makes sense in its current form, but I haven't looked at this for...

Update: - The cgroup approach bumped up against securiy concerns. - Putting on pause.

(spring cleaning the storage board) @bayandin is this ticket useful? I'm happy to close it if this is something we're unlikely to work on in the near future.

Currently, postgres_install.tar.gz is used when deploying bare-metal pageservers (we use the container image as a vehicle for distributing binaries). It is indeed just the same files that are already in...