Nathaniel Rutman <[EMAIL PROTECTED]> writes:
> Alastair McKinstry wrote:
>> Nathaniel Rutman wrote:
>>
>>> For what it's worth, ll_umount_begin is only called for clients, so
>>> maybe it is an issue with running a client on the same node as the
>>> MGS/MDT.
>>> Also, ll_umount_begin is only called in the umount -f case, not a
>>> non-forced case. So try the umount without the "-f".
>>>
>>>
>> Are you sure? I Just checked this, running 'umount /test' and oops'ed in
>> ll_umount_begin,
>> even though I did not do '-f'. This is in 2.6.18.
>>
>> - Alastair
>>
>>
> In 2.6.10 fs/namespace.c/do_umount():
> if( (flags&MNT_FORCE) && sb->s_op->umount_begin)
> sb->s_op->umount_begin(sb);
>
> Maybe it's changed in 2.6.18.
It has. The bugzilla has a patch.
MfG
Goswin
_______________________________________________
Lustre-discuss mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-discuss
