Unknown code adb 10 : cannot lock database error
ss488 at cornell.edu
Wed Nov 10 12:47:23 EST 2010
Thank you for the response. When we see these errors, usually large batches of Ids are being created or modified. Often it is not a problem, because processes try again and get through.
There was one instance when none of the processes were able to get through (I think due to a deadlock) and we had to restart the KDC. Are you doing a row level locking and is there a potential for a deadlock in the software ?
From: Ken Raeburn [mailto:raeburn at mit.edu]
Sent: Wednesday, November 10, 2010 1:56 AM
To: Shivakeshav Santi
Cc: krbdev at MIT.EDU
Subject: Re: Unknown code adb 10 : cannot lock database error
On Nov 9, 2010, at 14:12, Shivakeshav Santi wrote:
> We are seeing regular "Cannot lock database" errors on the KDC. The clients see "Unknown code adb 10" error.
>From a quick look at the sources, that can come up in a few cases. One of the database files may be locked already by another process; you can use lsof to see what other processes might have the files open. It may be trying to get an exclusive lock on a file it doesn't have write access to.
The attempts to lock the database files use non-blocking locking calls, and only make five attempts at one-second intervals before giving up. So if something else is keeping the database locked, either for more than five seconds, or for multiple briefer periods that happen to coincide with the attempts, that could be the problem. Do you know if anything else "interesting" is happening (KDB-related jobs running, or really intense load on the KDC or kadmin servers) at those times?
> We tried to retsrt the kadmin database and also run the following command but nothing seems to have helped.
> restorecon -R -v /var/kerberos/krb5kdc
'fraid I can't help with SELinux issues....
More information about the krbdev