Heikki Linnakangas 772d853dcf Fix race condition leading to panic in walkeeper. ...

The walkeeper launch two threads for each connection, and uses a guard
object to remove entry from 'replicas' array, when finishes. But only
the background thread held onto the guard object, so if the background
thread finished before the other thread, the array entry would be
removed prematurely, which lead to panic in the check_stop_streaming()
call.

Fixes https://github.com/zenithdb/zenith/issues/1103

2022-01-13 11:21:11 +02:00

..

bin

Add safekeeper --dump-control-file option.

2022-01-12 19:47:24 +03:00

http

Add term history to safekeepers.

2021-12-03 12:43:57 +03:00

callmemaybe.rs

Don't call immediately on resume in callmemaybe.

2022-01-03 20:44:36 +03:00

handler.rs

Parse cancel message in pq_proto (#1060)

2021-12-28 16:43:44 -05:00

json_ctrl.rs

Refactor Postgres command parsing in safekeeper.

2021-12-24 15:48:13 +03:00

lib.rs

Fix default safekeeper http port.

2022-01-11 10:13:27 +03:00

receive_wal.rs

send CallmeEvent::Unsubscribe request only when pageserver is caught up with safekeeper and it's time to stop streaming

2021-12-28 17:50:48 +03:00

s3_offload.rs

Switch safekeeper from log to tracing logging.

2021-12-09 06:57:46 +03:00

safekeeper.rs

Add safekeeper --dump-control-file option.

2022-01-12 19:47:24 +03:00

send_wal.rs

Fix race condition leading to panic in walkeeper.

2022-01-13 11:21:11 +02:00

timeline.rs

Fix race condition leading to panic in walkeeper.

2022-01-13 11:21:11 +02:00

upgrade.rs

Add safekeeper --dump-control-file option.

2022-01-12 19:47:24 +03:00

wal_service.rs

Make safekeeper namings more consistent with reality.

2021-12-21 13:24:23 +03:00