The AI-native database built for LLM applications, providing incredibly fast hybrid search of dense vector, sparse vector, tensor (multi-vector), and full-text
Hardware: X86_64
OS type:
$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 22.04.4 LTS
Release: 22.04
Codename: jammy
$ uname -a
Core Info: Linux ibp 5.15.0-122-generic #132-Ubuntu SMP Thu Aug 29 13:45:52 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux
infinity version:
Release: 0.5.0 build on 2024-11-05 13:21.45 with RelWithDebInfo mode from branch: main, commit-id: bcde5251945c88670a5f09c98c1c10d11e05affc
Actual behavior and How to reproduce it
Configure a cluster with one leader, one follower, and one learner.
Directly kill the learner using the kill -9 command, and the leader and follower cannot perceive it.
When viewing on the leader and follower nodes, the learner's status remains 'alive', but the heartbeat count has not changed.
Expected behavior
The expectation is that after killing the learner for a certain period of time, the other nodes should see information about the learner node being down when checking the cluster status.
Additional information
The unperceived state remains unchanged until the leader node modifies the log, at which point the leader's downtime will be detected. This will result in a mismatch between the states of the leader and the follower nodes.
Is there an existing issue for the same bug?
Version or Commit ID
dfe9caf19a59b73406817d393f1a24e97dde81a4
Other environment information
Actual behavior and How to reproduce it
Configure a cluster with one leader, one follower, and one learner. Directly kill the learner using the kill -9 command, and the leader and follower cannot perceive it.
When viewing on the leader and follower nodes, the learner's status remains 'alive', but the heartbeat count has not changed.
Expected behavior
The expectation is that after killing the learner for a certain period of time, the other nodes should see information about the learner node being down when checking the cluster status.
Additional information
The unperceived state remains unchanged until the leader node modifies the log, at which point the leader's downtime will be detected. This will result in a mismatch between the states of the leader and the follower nodes.
No response