prometheus

mirror of https://github.com/prometheus/prometheus.git synced 2025-09-21 13:51:00 +02:00

History

fix: storage/remote.pool interned refs count and flaky test (#16335 )

* Fix storage/remote.pool interned refs count and flaky test

I saw TestIntern_MultiRef_Concurrent failing on a different PR saying 'expected refs to be 1 but it was 2'.

I took a look, and it definitely can be racy, especially with a time.Sleep() of just 1ms.

I'm fixing that by explicitly waiting until it has been released, and by repeating that 1000 times, otherwise it's just a recipe for a future flaky test.

OTOH, I also took a look at the implementation and saw that we were not holding the RLock() when increasing the references count, so when releasing there was a race condition for the cleanup, I fixed that by holding RLock() while increasing the references count.

Signed-off-by: Oleg Zaytsev <mail@olegzaytsev.com>

* s/Equalf/Equal/

Signed-off-by: Oleg Zaytsev <mail@olegzaytsev.com>

---------

Signed-off-by: Oleg Zaytsev <mail@olegzaytsev.com>

2025-03-27 21:04:30 +01:00

remote

fix: storage/remote.pool interned refs count and flaky test (#16335 )

2025-03-27 21:04:30 +01:00

buffer_test.go

Unit test for regression in rate vs float and histogram mixup

2024-10-02 13:56:33 +02:00

buffer.go

chore: fix function name in comment

2024-10-29 17:58:44 +08:00

errors_test.go

Fix appendable: check whether last val was a histogram (#14613 )