Endless loop in logging

[tomek@…]

Ocasionally when I stop (Ctrl-C) tracd, I goes into endless exception loop:

[...]
Traceback (most recent call last):
  File "/usr/lib/python2.4/logging/__init__.py", line 737, in emit
    self.stream.write(fs % msg)
ValueError: I/O operation on closed file
Traceback (most recent call last):
  File "/usr/lib/python2.4/logging/__init__.py", line 737, in emit
    self.stream.write(fs % msg)
ValueError: I/O operation on closed file
[...]

The only way to stop Trac then is SIGKILL its process.

[anonymous]

What version are you using? Might be important.

[tomek@…]

The version I use is 0.11dev-r6372

[Christian Boos]

Milestone 1.0 deleted

[Christian Boos]

See ticket:8443#comment:37 - if in some conditions this error (or similar) happens for a thread in httpd or trac.fcgi, search no further for the pathological situations where we observer 100% CPU usage for extended periods of time…

The point is that we do lots of things in our error handlers, and if one of those things breaks, then all sorts of weird things can happen (e.g. r9686).

[Christian Boos]

I believe r9713 should fix this issue.

It remains to be seen if this was really the reason for the 100% CPU problem. Also, the change could be made more robust. In particular, I currently rely on the fact that the only call to env.shutdown() is done within a critical section.

Keeping the issue on the radar and opened for a few more days.

[Christian Boos]

Ok, so r9713 probably fixed the issue reported here, so I'm closing the ticket (please reopen if nevertheless a similar error re-appears).

This didn't help for #8443, though.