John Spray

Thanks for the ping. Can we enforce a size limit of this WAL write on the postgres side? (i.e. check `size` in wallog_file_descriptor, and if it exceeds a threshold, log...

To factor in the issues we've had with capacity/replay scaling, what do you think about having separate stores for "big values" (such as pgstats) and "small values" (such as logical...

Cross referencing with https://github.com/neondatabase/neon/pull/6874#pullrequestreview-1901765234 -- since that PR is writing image values regularly for the AuxFilesDirectory structure, it's important that we don't put the multi-megabyte stats value in the same...

> So even if compute is restarted each 5 minutes, and there are 10k tables, still it could not cause storage bloating. It doesn't matter how often the compute is...

- Action this week: benchmark changes and progress PRs.

Merging these changes will block on writing an improved benchmark

Ongoing: gradual rollout of sync/async configurable code + switching on the async mode. EOW goal: deploy across staging (a prod region next week)

Same plan this week: config change in preprod after the release, then gradually in prod.

> API (draft) that I'm proposing is consisting of two parts: >GET /abandoned The "abandoned" timeline can have many children, so the pageserver needs to know which one to merge...

Notes from chat: - The Console concept is to "abandon" a timeline: this means that a user-facing Branch no longer refers to the timeline. - In the RFC example (child...