Thanks for all the info. Could you create a new bug on our Bugzilla
(http://bugs.gluster.com/) and paste the same info there too? That
way we can keep track of the issue if it ever pops up in our testing
and you can get an update when it is resolved.
I will. For what it's worth, I just had it reoccur with readahead commented out. Given that it might happen again out of office hours when I can't quickly resolve it I need to take out of production rotation in an hour or so, so I won't be able to do much more testing today - it took more than 5 hours to reproduce this time.
Regards,
Vidar