We are running a two node NFS failover cluster under Scientific Linux 4.7, which is based on the Red Had distribution.
Kernel is 2.6.9-42.0.3.ELsmp on both nodes.

I notice, that on node "A" the file /proc/locks has more than 500000 entries, where as node "B"only lists 6 entries (cluster daemon, fence daemon, sendmail, ...).
As far as I can see, all entries in /proc/locks look identically on node "A":

---snip---8<----
488: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
489: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
490: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
491: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
492: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
493: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
494: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
495: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
496: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
497: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
498: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
499: POSIX ADVISORY WRITE 11914 fc:04:472130 0 EOF
---snip---8<----

I assume this is not meant to be like this, but I cannot find a process number 11914!

Can anyone tell me, how to find a file based on its Major:Minor:Inode (fc:04:472130)?
Has anyone seen this before?
I appreciate any hint/help/suggestion!
Thanks - Rene Kapeller