feat(replication): fix seg fault when ack is received after FeedSlaveThread::Stop is called #3101
Conversation
| warn("Slave thread was terminated, would stop feeding the slave: {}", conn_->GetAddr()); | ||
| } | ||
|
|
||
| void FeedSlaveThread::Join() { | ||
| if (auto s = util::ThreadJoin(t_); !s) { | ||
| warn("Slave thread operation failed: {}", s.Msg()); | ||
| } else { | ||
| // after the loop is finished, the connection can be freed | ||
| conn_.reset(); |
There was a problem hiding this comment.
I do not find where the connection is released after FeedSlaveThread stops. Please let me know if this should not be done.
There was a problem hiding this comment.
Yes, it should be correct to reset the unique pointer only here. Because the connection has been detached from the worker before using it in FeedSlaveThread, so it's nothing to be done.
| @@ -622,3 +625,126 @@ func TestSlaveLostMaster(t *testing.T) { | |||
| duration := time.Since(start) | |||
| require.Less(t, duration, time.Second*6) | |||
| } | |||
|
|
|||
| func TestWALDiscreteError(t *testing.T) { | |||
There was a problem hiding this comment.
I managed to create a WAL with discrete sequence number to trigger the error, but I cannot trigger the seg fault. I think it is because the seg fault only happens when ack is sent back on the same connection after FeedSlaveThread::Stop call and the time window is small.
I am leaving it here because it does no harm to test an extra case, but let me know if there is better way.
There was a problem hiding this comment.
nvm, this is not related to the seg fault
zhixinwen
force-pushed
the
zhixin/fix-seg-fault
branch
5 times, most recently
from
6c89761 to
2da98d2
Compare
zhixinwen
force-pushed
the
zhixin/fix-seg-fault
branch
from
2da98d2 to
95ecf41
Compare
fix: #3097