Folks,
I have a system that has a sporadic problem, and I can't seem to find any
info on Google about it.
I am running a variant of 2.4.9-34. A certain process is locking access to
a filesystem, so that any subsequent access to it hangs the shell. I'm
seeing:
#ps xwo pid,command,wchan |grep inode
21062 <program name> wait_on_inode
Looking at the source code, I can see that the inode attribute I_LOCK is
getting set, but for some reason is not being cleared. At the very least,
I'd like to know *which* inode is being locked so I can narrow the problem
down to an application developed internally and a third-party database.
I've tried debugfs and lde, but I can't get either to indicate a locked
inode. I also used a kernel patch from Bill Hawes from '98 to dump inode
info, but it oopses, probably due to it being 5 years out of date
(http://tinyurl.com/srir)
Does anyone have a handy utility/patch that can help pinpoint the errant
inode?
Thanks
Mark
