Christian Schwarz ce0d0a204c fix(walredo): shutdown can complete too early (#8701) ...

Problem
-------

The following race is possible today:

```
walredo_extraordinary_shutdown_thread: shutdown gets until Poll::Pending of self.launched_processes.close().await call

other thread: drops the last Arc<Process>
  = 1. drop(_launched_processes_guard) runs, this ...

walredo_extraordinary_shutdown_thread: ... wakes self.launched_processes.close().await

walredo_extraordinary_shutdown_thread: logs `done`

other thread:
  = 2. drop(process): this kill & waits
```

Solution
--------

Change drop order so that `process` gets dropped first.

Context
-------


https://neondb.slack.com/archives/C06Q661FA4C/p1723478188785719?thread_ts=1723456706.465789&cid=C06Q661FA4C

refs https://github.com/neondatabase/neon/pull/8572
refs https://github.com/neondatabase/cloud/issues/11387

2024-08-12 18:15:48 +01:00

..

benches

pageserver: make bench_ingest build (but panic) on macOS (#8641)

2024-08-07 21:17:08 +01:00

client

feat(storcon): timeline detach ancestor passthrough (#8353)

2024-07-15 18:08:24 +03:00

compaction

Fix nightly warnings 2024 june (#8151)

2024-07-12 13:58:04 +01:00

ctl

Use DefaultCredentialsChain AWS authentication in remote_storage (#8440)

2024-07-19 21:19:30 +02:00

pagebench

Move keyspace utils to inherent impls (#7929)

2024-06-03 16:18:07 +02:00

src

fix(walredo): shutdown can complete too early (#8701)

2024-08-12 18:15:48 +01:00

test_data

pageserver: add layer visibility calculation (#8511)

2024-08-01 09:25:35 +00:00

Cargo.toml

pageserver: add bench_ingest (#7409)

2024-08-06 16:39:40 +00:00