[
https://issues.apache.org/jira/browse/RATIS-2507?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sammi Chen updated RATIS-2507:
------------------------------
Description:
RaftLog end log index is smaller than last snapshot index during RaftServer
startup, it indicates that raft log state and state machine state is
inconsistent.
In this case, it's better to fail the RaftServer, instead of currently log a
WARN message and continue.
A real case, is an OM failed to install snapshot and terminate itself. The on
disk state machine state(rocksdb) and raft log files are from different source,
state machine data from leader, raft log files are local, and snapshot index
get from state machine is bigger than raft log end index, which further cause
another issue.
Refer to https://issues.apache.org/jira/browse/HDDS-15103 to get more info.
was:
RaftLog end log index is smaller than last snapshot index during RaftServer
startup, it indicates that raft log state and state machine state is
inconsistent.
In this case, it's better to terminate the RaftServer, instead of currently log
a WARN message and continue.
A real case, is an OM failed to install snapshot and terminate itself. The on
disk state machine state(rocksdb) and raft log files are from different source,
state machine data from leader, raft log files are local, and snapshot index
get from state machine is bigger than raft log end index, which further cause
another issue.
Refer to https://issues.apache.org/jira/browse/HDDS-15103 to get more info.
> Fail RatisServer when RaftLog end log index is smaller than last snapshot
> index during startup
> ----------------------------------------------------------------------------------------------
>
> Key: RATIS-2507
> URL: https://issues.apache.org/jira/browse/RATIS-2507
> Project: Ratis
> Issue Type: Bug
> Reporter: Sammi Chen
> Assignee: Sammi Chen
> Priority: Major
>
> RaftLog end log index is smaller than last snapshot index during RaftServer
> startup, it indicates that raft log state and state machine state is
> inconsistent.
> In this case, it's better to fail the RaftServer, instead of currently log a
> WARN message and continue.
> A real case, is an OM failed to install snapshot and terminate itself. The on
> disk state machine state(rocksdb) and raft log files are from different
> source, state machine data from leader, raft log files are local, and
> snapshot index get from state machine is bigger than raft log end index,
> which further cause another issue.
> Refer to https://issues.apache.org/jira/browse/HDDS-15103 to get more info.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
