Fix intermitent condition where onerror is called when conn is closed by client#72
Fix intermitent condition where onerror is called when conn is closed by client#72ngbrown wants to merge 1 commit intocloudamqp:mainfrom
Conversation
3e95f50 to
cb48bbe
Compare
|
Nice, this gets rid of the |
|
There should be an option to allow maintainers to push to your branches/PRs, then I can update (rebase) and merge them, so you don't have to do all that work, and will look like they was merged too. :) |
I've seen that option on Pull Requests in GitHub in the past, but it's not there on these PRs. It might be because I'm forking into an separate account/org? |
I saw it working for another PR, it said (in the right side-bar)
Sounds like that reading https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/allowing-changes-to-a-pull-request-branch-created-from-a-fork
|
The test "onerror is not called when conn is closed by client" intermittently fails.
According to some resources, such as RFC-1122 (4.2.2.12 Closing a Connection): if the host actively closes a connection, while still having unread incoming data available, the host should send the signal
RST(losing any received data) instead ofFIN. This may or may not be what is happening, but there's there's something causing aRSTto be received by the client at least some of the time.This pull request proposes the solution of calling
socket.destroy()incloseSocket()to ignore the possibility of aRSTresponse. This was also the solution in postwait/node-amqp#444.See recent past failures: