You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First log error on pageserver_2: (1603, '2024-12-09T17:12:30.404054Z WARN request{method=PUT path=/v1/tenant/dc41c9c7a8d2d4046d786f9e05a86d1e/location_config request_id=ee51f4e1-5bc3-4f62-be8c-b0f0ecedcea1}:upsert_location{tenant_id=dc41c9c7a8d2d4046d786f9e05a86d1e shard_id=0000}: offloaded timeline dd17b03eb917223858a7dfcd5dd0cc68 was dropped without having cleaned it up at the ancestor\n')
Hint: use scripts/check_allowed_errors.sh to test any new allowed_error you add
The text was updated successfully, but these errors were encountered:
Looking at the logs, this seems to be offload_timeline racing with Tenant::shutdown, in this case, with upsert_location during an Attached->Secondary transition.
A : compaction task decides to offload timeline
B: tenant shutdown is called
B: tenant defuses for drop all of the offloaded timelines
B waits for timeline shutdowns to complete
A: compaction task removes the timeline from timelines and adds offloaded timeline to tenant. finishes with the offload.
B finished waiting, drops the Tenant object
The issue is that we add the offloaded timeline to the tenant after the defuse_for_drop loop of shutdown has already ran.
https://neon-github-public-dev.s3.amazonaws.com/reports/pr-9978/12240079195/index.html#testresult/314b8921da5c5a9/retries
The text was updated successfully, but these errors were encountered: