John Spray

Moved out of storage team tasks board (no problem to keep a draft refactor PR around, but we don't need to track it there).

Idea: we could combine this with handling of idle tenants in the rest of the stack, e.g. for an idle tenant pageserver would like to be woken up to GC...

Status: https://github.com/neondatabase/neon/issues/6338 is blocked on https://github.com/neondatabase/neon/issues/8542

Status 2024-01-29: - Python script is written, hasn't been tested in the field yet. Will be used today in staging. - Change to delete initdbs has merged - Doc is...

Update: - Access control issues when testing, which may be resolved: need to check - This week try again to validate procedure.

Triage notes: - Christian's latest update is current

Changes that upload remote index proactively in some cases -- this was for sharding, might not covery every case where a timeline ingest goes idle https://github.com/neondatabase/neon/blob/221414de4b0260056e0961528d46c5141825a0a0/pageserver/src/tenant/timeline.rs#L1202-L1226

Currently advancing the disk consistent lsn is behind a condition for whether we're sharded, which can be safely removed (the condition was just to de-risk the change): https://github.com/neondatabase/neon/commit/ac7fc6110bba250f17b494c604b717cf69e09ef1

On second thoughts, we might not have a problem here: `Timeline::collect_keyspace` is generating keyspace ranges from relation start block + size, so it will generate ranges that cover all the...

Folding this into https://github.com/neondatabase/neon/issues/6774