[test-failure] Slot-migration related

[roshkhatri]

*** Migration source is auto-updated after failover in source shard in tests/unit/cluster/slot-migration.tcl
*** Replica redirects key access in migrating slots in tests/unit/cluster/slot-migration.tcl
*** Replica of migrating node returns ASK redirect after READONLY in tests/unit/cluster/slot-migration.tcl
*** Replica of migrating node returns TRYAGAIN after READONLY in tests/unit/cluster/slot-migration.tcl
*** Replica of importing node returns TRYAGAIN after READONLY and ASKING in tests/unit/cluster/slot-migration.tcl
*** New replica inherits migrating slot in tests/unit/cluster/slot-migration.tcl
*** New replica inherits importing slot in tests/unit/cluster/slot-migration.tcl

These tests seems to be consistently failing in the daily runs.

I tried to narrow down the commit #2301 that might possibly cause this.

I have tested this by reverting this commit on my personal repo and found the tests not failing

[roshkhatri]

@enjoy-binbin can you please take a look?

[enjoy-binbin]

oh, i did not realize it affect the slot migration, i will take a look when i have time (maybe this weekend)

[roshkhatri]

I think its a timing issue on test-ubuntu-lttng and test-ubuntu-jemalloc-fortify

[roshkhatri]

Seems like increasing the wait just enough solved for fortify however now for lttng https://github.com/roshkhatri/valkey/actions/runs/16382825913/job/46297474597

[enjoy-binbin]

i don't think the wait will help, i am taking a look, i have some clues.

[enjoy-binbin]

@PingXie I think the reason is that, becasue in #2301 we call clusterMoveNodeSlots to move the slots, and now we dont have the chance to call clusterUpdateSlotsConfigWith since we updated the new primary slots.

wait_for_slot_state 1 "\[609-<-$R3_id\]"
23537:M 19 Jul 2025 00:29:23.382 * Failover occurred in migration source. Update importing source for slot 609 to node 4621ba77ff67c661445c8e1b7f066ba19bc993a1 () in shard 465fc201a2d23de28f7664270ceb2f0b024670c8.

wait_for_slot_state 1 "\[609-<-$R0_id\]" ---- fail
23537:M 19 Jul 2025 00:29:24.866 * A failover occurred in shard 465fc201a2d23de28f7664270ceb2f0b024670c8; node 4621ba77ff67c661445c8e1b7f066ba19bc993a1 () lost 5462 slot(s) and failed over to node 5f38436c7510a26cf7561259265c54fc38c1f30b () with a config epoch of 10

So we should also move the importing and migrating slots in clusterMoveNodeSlots?

[PingXie]

So we should also move the importing and migrating slots in clusterMoveNodeSlots?

make sense but now I think I will need to take another look at #2370