Overview
reshard_timeseries_building_index_failover.js and reshard_timeseries_failover_during_abort.js (added in SERVER-126218) are failing in the sharding_last_lts multiversion suite with:
BadValue: Chunk range must start at global min for new shard key
Root Cause
The tests only use stepUpNewPrimaryOnShard (no binary restart), so the multiversion_incompatible tag is not appropriate. The underlying root cause is that the v8.0 backport of SERVER-111411 (BACKPORT-26846) has not yet landed, meaning mixed-version clusters still exhibit the old behavior that triggers this error during timeseries resharding failover scenarios.
Fix
Add both tests to etc/backports_required_for_multiversion_tests.yml as a temporary exclusion until BACKPORT-26846 (v8.0 backport of SERVER-111411) lands. Once the backport merges, these entries should be removed.
Affected Tests
- jstests/sharding/reshard_timeseries_building_index_failover.js
- jstests/sharding/reshard_timeseries_failover_during_abort.js
References
- SERVER-126218 — original ticket that added these tests
- SERVER-111411 — upstream fix (closed)
- BACKPORT-26846 — v8.0 backport (pending)
- SERVER-127893 — full root cause analysis
- BF-43571 — BF for reshard_timeseries_building_index_failover.js
- BF-43572 — BF for reshard_timeseries_failover_during_abort.js
- is related to
-
SERVER-111411 Resharding Does Not Properly Handle forceRedistribution With Time Series Collections
-
- Closed
-
-
SERVER-127893 Investigate why resharding timeseries abort tests fail with BadValue in sharding_last_lts multiversion context
-
- Needs Scheduling
-