shinrich commented on pull request #7134: URL: https://github.com/apache/trafficserver/pull/7134#issuecomment-680187419
I just pushed a less extreme change. Rather than removing the inactivity logic completely from write_disable/read_disable, I reverted how the next_inactivity_timeout_at was cleared to 0 that variable directly rather than calling set_inactivity_timeout() which will also clear inactivity_timeout_in. The call to set_inactivity_timeout() was added in the NetEvent restructuring I think. Our ATS7 and the code @oknet references identifies the code that directly clears only net_inactivity_timeout_at. I verified that this code works for my tunnel timeout case too. UnixNetVConnection::set_enabled is called well before inactivity_cop triggers, so the original inactivity_timeout is used to recompute next_inactivity_timeout_in. I think UnixNetVConnection::set_default_inactivity_timeout should also be updated to not clear inactivity_timeout_in, but I will leave that to another PR. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
