Snapshot Segment Mismatch

[fredfortier]

I'm aware that this is likely an implementation issue on my part, but do you have any hint on what could explain an error like this when installing a snapshot.

Failed to install snapshot "snapshot segment id mismatch, expect: snapshot-257+0, got: snapshot-257+39845888"

I haven't reproduced this consistently yet, but I'm working on it. I've this error with earlier versions of async_raft. I know it happens with large snapshots but I don't know yet whether there's a threshold. Any pointers would be appreciated.

[github-actions[bot]]

👋 Thanks for opening this issue!

Get help or engage by:

• /help : to print help messages.
• /assignme : to assign this issue to you.

[drmingdrmer]

It looks like there is a retransmit of a snapshot segment.

The follower/learner has finished installing a snapshot and has cleaned up the session by setting RaftCore.snapshot_state to None.

But the leader did not wait long enough for the follower/learner to respond. Then the leader retransmitted a segment.

To confirm that, you need to check the debug level log to see if there are a retransmitted snapshot segment. The debug logging statement is:

            // Send the RPC over to the target.
            tracing::debug!(
                snapshot_size = req.data.len(),
                req.offset,
                end,
                req.done,
                "sending snapshot chunk"
            );