Christian Schwarz

Bug triage meeting: - tactical fix: wrap it retry loop with infinite retries; must remain sensitive to location_config detach etc - long-term: make attach synchronous / poll for completion from...

``` error[E0583]: file not found for module `entry` --> libs/neon-shmem/src/hash.rs:17:1 | 17 | pub mod entry; ```

Removing myself from review, this change is inside pgxn and doesn't require storage review afaict.

Removing myself from review, speaking as a storage person, I don't feel qualified to review this. If there's anything in the _design_ that you want reviewed / cross-checked with storage...

Hey @ding-young , sure, give it a try. It's probably some `tokio::select!` that stops running the size calculation future to completion. Size calculation _should_ be sensitive to Timeline::cancel/Tenant::cancel nowadays, so,...

@VladLazar found another occurence in - https://github.com/neondatabase/neon/issues/12006 > I noticed a bunch of these last week: `dropping ValuesReconstructState while some IOs have not been completed`. Full log example [here](https://neonprod.grafana.net/goto/ZHFNnzfNR?orgId=1). I...

Vlad proposes here to just skip step 3 and go straight to `node-delete`. - https://github.com/neondatabase/cloud/issues/20895#issuecomment-2729556653 This avoids the problem with `conf = None` that was pointed out in the "Background"...

We had the call, I now understand the rationale for not failing `LayerManager::layers()` and callers after we've shut down the `LayerManager`. I really like this impl, will do more review...

> In future, what we really need is some ActiveTimeline guard that holds a gate and ensures the interior timeline is not shut down, so that it can do infallible...

This week: * last touches on https://github.com/neondatabase/neon/pull/8537 * second review from John * merge by Wednesday morning * soak in staging until next week Next week: * perf qualification in...