Description of problem: <Mar/29 02:38 pm>VFS: Busy inodes after unmount of 0:17. Self-destruct in 5 seconds. Have a nice day... <Mar/29 02:38 pm>Unable to handle kernel NULL pointer dereference at 0000000000000010 RIP: Version-Release 2.6.18-8.1.3.el5 How reproducible: No reliable reproducer. The error has shown up once while running the RHTS connectathon tests.
Hmm with no reliable reproducer, this might be tough to track down. I'll have a look over the oops and see what we might be able to determine from it (but I fear that the answer there is "not much").
Yes, it's a tough one to reproduce. Factors that tend to correlate with it are nfs4, autofs (use short timeout to remount the file system often) and, I get impression, several user accessing the same filesystem (like in: first one user triggers the mount, next, another one). I have seen "Busy inodes" with 2.6.18-8.1.8 too, although it has clearly become more difficult to trigger.
Its been quite a while since this has been seen, so I'm going to close this as INSUFFICIENT_DATA. If this oops happens again, please reopen...