Steve Mcclure wrote:
>
> (I sent this so the NFS lists too since it's an autofs/NFS specific
> interaction that I'm having.)
>
> I remember mention of a VFS layer bug which affects autofs (among
> other things) in kernel versions 2.2.x and 2.3.x, and which had been
> fixed in the 2.3.x kernels (somewhere around 2.3.14 or 2.3.16 or so)
> and that it would possibly be backported to 2.2.x.
>
> Since I believe that I'm being bitten by this bug (which I'll describe
> below) I was wondering if there was any word of the status of the VFS
> fix for 2.2.x kernels? (Since I require a 2.2.x kernel, it would be
> nice for it to not hang on me when my network goes belly up for a few
> seconds at a critical time. :) )
>
> The bug I see is that autofs will hang and require a reboot of the
> client if the client receives RPC timeouts or the NFS server not
> responding during the automount of an NFS filesystem. There was a
> patch out there that was an improper workaround (which could cause
> other problems) for the VFS bug, and applying that patch "kinda" fixes
> this specific problem - the client no longer requires a reboot, but
> you need to stop the user level automouters, unmount all automounted
> stuff by hand and restart the automounters, and then everything is
> fine again (until it happens again.)
>
> Thanks!
>
> -- Steve McClure
> [EMAIL PROTECTED]
I have asked Linus to forward this fix to Alan. I don't know if he has,
yet.
Be careful with the "patch" - it may cause silent data corruption, I
believe.
-hpa
