failed replication_test.py::test_replication_timeout_on_full_sync

[BorysTheDev]

https://github.com/dragonflydb/dragonfly/actions/runs/13892743446/job/38867312517

[BorysTheDev]

somehow we are stuck on the

  VLOG(1) << "Waiting for all full sync cut confirmations";
  sync_block->Wait();

next messages were done after terminating because sync_block->cancel() was called in err_handler

[BorysTheDev]

Possible cause: to cancel replication we use Cancel error, but the streamer in this case can skip the callback WaitForInflightToComplete check

void JournalStreamer::Cancel() {
  ...
  if (!cntx_->IsError()) {
    WaitForInflightToComplete();
  }
}

[kostasrim]

Possible cause: to cancel replication we use Cancel error, but the streamer in this case can skip the callback WaitForInflightToComplete check

void JournalStreamer::Cancel() {
  ...
  if (!cntx_->IsError()) {
    WaitForInflightToComplete();
  }
}

yes already aware of this. If you recall I was checking memory ordering around that synchronization primitive but IMO I think there is another issue.

[BorysTheDev]

yes already aware of this. If you recall I was checking memory ordering around that synchronization primitive but IMO I think there is another issue.

Agree. I just write about it to not forgetting in the future