[Lustre-discuss] Failure to umount filesystem complete

[springmelco11321]

We are running Lustre-2.1.2 on Redhat6.2, and using Lustre-1.8.7 on Redhat4.8 as client. 
On client we use autofs to mount and umount the filesystem, recently the filesystem cannot umount complete frequently because of network problem.
when this happens, we have to reboot the client, Is another way to deal with that?

More info as below:
[root at client1 ~]#lfs check mds
testfs-MDT0000-mdc-00000100c55ac400: INACTIVE 
[root at client1 ~]#lctl dl
1 UP lov testfs-clilov-00000100c55ac400 8acad536-bd01-0204-08f2-3a804613a686 3
2 UP mdc testfs-MDT0000-mdc-00000100c55ac400 8acad536-bd01-0204-08f2-3a804613a686 5

and when I use the commond to get the state of mdc (lctl get_param mdc.*.*.state), the commond becomes pending.
when using "umount -f ", there is no responding also.

how can we deal with this situation except reboot?

Thanks in advance for any help offered.




springmelco11321
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20130930/7973ace3/attachment.htm>