Node votes non-Nil for a proposal it has not received

[CHr15F0x]

2025-11-28T20:48:59.0356252Z       2025-11-28T20:47:33 ERROR pathfinder: Consensus P2P event processing task ended unexpectedly, task_result: Err(No ValidatorStage for height and round (14:0))
2025-11-28T20:48:59.0356379Z         at crates/pathfinder/src/bin/pathfinder/main.rs:948

2025-11-28T20_33_59_0704256Z.log

https://productionresultssa1.blob.core.windows.net/actions-results/eaac18e6-5bfd-4bf0-b4f9-ab398d5b8e5b/workflow-job-run-9cc5e009-cb8a-53f9-b84b-484660e92d70/logs/job/job-logs.txt?rsct=text%2Fplain&se=2025-11-28T21%3A01%3A13Z&sig=VEB6DfWKvWE1OEBkUobL%2FvLljKgqoOM7roSyHfxr8qM%3D&ske=2025-11-29T07%3A26%3A09Z&skoid=ca7593d4-ee42-46cd-af88-8b886a2f84eb&sks=b&skt=2025-11-28T19%3A26%3A09Z&sktid=398a6654-997b-47e9-b12b-9515b896b4de&skv=2025-11-05&sp=r&spr=https&sr=b&st=2025-11-28T20%3A51%3A08Z&sv=2025-11-05

[CHr15F0x]

I cannot reproduce this locally (yet).

[CHr15F0x]

The failure is related to the fact that when Bob has recovered after H=13 is finalized, he misses the proposal for H=14, co that proposal is then not cached and persisted. Since, after respawning, Bob has missed all the parts of the proposal @h=14, there is no validator present in the cache. In this scenario Bob should vote NIL on H=14 but instead he sent valid votes.

Bob voting NIL @h=14 will cause the other nodes to move to the next round and re-propose in a 3 node scenario. If the number of nodes is >3, Bob should fall behind and use catch up sync.

[CHr15F0x]

Proposal for H=14 has been received, however recovery does not reconstruct the block validator. Adding this feature should solve the problem.