Bug 852195

Created attachment 607342 [details]
Core dump of saslauthd stuck in poll loop

Description of problem:

Saslauthd periodically ends up in a tight loop inside krb5_get_credentials, with the kerberos libraries continuously calling poll and getting an error on one of the file descriptors back. lsof is unable to decode what the file descriptor is. Eventually all saslauthd processes are deadlocked, and nothing can authenticate. When strace'd, a saslauthd process is just doing this:

poll([{fd=9, events=POLLIN}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLOUT}], 3, -5521820) = 1 ([{fd=12, revents=POLLIN|POLLERR|POLLHUP}])
poll([{fd=9, events=POLLIN}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLOUT}], 3, -5521820) = 1 ([{fd=12, revents=POLLIN|POLLERR|POLLHUP}])
poll([{fd=9, events=POLLIN}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLOUT}], 3, -5521820) = 1 ([{fd=12, revents=POLLIN|POLLERR|POLLHUP}])
poll([{fd=9, events=POLLIN}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLOUT}], 3, -5521820) = 1 ([{fd=12, revents=POLLIN|POLLERR|POLLHUP}])
poll([{fd=9, events=POLLIN}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLOUT}], 3, -5521820) = 1 ([{fd=12, revents=POLLIN|POLLERR|POLLHUP}])

Version-Release number of selected component (if applicable):

krb5-libs-1.6.1-70.el5
cyrus-sasl-2.1.22-5.el5_4.3

How reproducible:

This occurs periodically as our users authenticate, but we have not found the trigger that causes the failure. It can take as long as a few hours or as short as a few minutes for the problem to manifest. 

Since we cannot find the cause, the next few questions are unanswered.

Steps to Reproduce:
1.
2.
3.
  
Actual results:


Expected results:


Additional info:

Attached is the core dump of a saslauthd process with the issue.