NFS and/or autofs hanging on Linux

NFS and/or autofs hanging on Linux

[Burke, Alan]

Hi Everyone,

We have got a problem with NFS hanging with automounter mounts.

We're running RH8 2.4.20-20.0 kernel
NFS-utils 1.0.1-2
Autofs4.1.2 (we must have the /net ability)
H/W is mixed HP4100's, Dell 2650's, Sun V60 and V65's
All the machines are in a Sun NIS domain with 700 Sun machines that run
fine.

The problem is intermittent; any automounter point can hang at any time.
For example auto_home may have 1000 entries
cd /home/user1 hangs and /home/user2 does not on machine Foo, but
/home/user2 is fine on machine Bar.

Once hung we sometimes can get it to release by umount -at nfs others
time a reboot is the only way to get it to release.

any help is appreciated

thanks,

alan



	 =20





-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: NFS and/or autofs hanging on Linux

[raven]

On Thu, 7 Oct 2004, Burke, Alan wrote:

> Hi Everyone,
> 
> We have got a problem with NFS hanging with automounter mounts.
> 
> We're running RH8 2.4.20-20.0 kernel
> NFS-utils 1.0.1-2
> Autofs4.1.2 (we must have the /net ability)
> H/W is mixed HP4100's, Dell 2650's, Sun V60 and V65's
> All the machines are in a Sun NIS domain with 700 Sun machines that run
> fine.

Have you patched the kernel with the latest autofs4 patches from 
kernel.org?

You might like to visit Jeff Moyers people page at RedHat to get his 
latest RPMS.

http://people.redhat.com/jmoyer

Ian



-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

force unmount not working 2.4.22

[Goutham Kurra]

Force unmount simply isn't working on a standard linux
2.4.22 kernel. Isn't it supposed to unmount NFS
filesystems no matter what?

In my case, the backend server is still up and
working, I have a "tail -f /mnt/nfs/foo" of some file
on the mount, and "umount -f" fails with "device or
resource is busy"

I notice that rpc_killall_tasks() is called from the
umount_begin(), but none of the processes get an EIO.
They continue to work as before.

I've also noticed this on stock RH 2.4.20-9 kernels.

Any idea what's going on?

Thanks!
g



		
__________________________________
Do you Yahoo!?
Take Yahoo! Mail with you! Get it on your mobile phone.
http://mobile.yahoo.com/maildemo 


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: force unmount not working 2.4.22

[Garrick Staples]

[-- Attachment #1: Type: text/plain, Size: 892 bytes --]

On Sat, Oct 09, 2004 at 02:22:31PM -0700, Goutham Kurra alleged:
> 
> Force unmount simply isn't working on a standard linux
> 2.4.22 kernel. Isn't it supposed to unmount NFS
> filesystems no matter what?
> 
> In my case, the backend server is still up and
> working, I have a "tail -f /mnt/nfs/foo" of some file
> on the mount, and "umount -f" fails with "device or
> resource is busy"

I don't believe force is useful in this case, what you want is -l.

   -f     Force unmount (in case of an unreachable NFS system).  (Requires
          kernel 2.1.116 or later.)

   -l     Lazy unmount. Detach the filesystem from the filesystem  hierar-
          chy now, and cleanup all references to the filesystem as soon as
          it is not busy anymore.  (Requires kernel 2.4.11 or later.)

-- 
Garrick Staples, Linux/HPCC Administrator
University of Southern California

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: force unmount not working 2.4.22

[Goutham Kurra]

I can see that '-f' tries to stop all active rpc tasks
to the server, but I guess the reference counts on the
mount are not dropped. What would be cool is a
'brute-force' option that kills any process that still
holds a reference to the filesystem and then also
unmounts it. 

The problem with using something like fuser is that it
will promptly hang on an unreachable server making it
useless on nfs filesystems. 

Any ideas on what to do in this case (i.e, unreachable
NFS server, plus a bunch of processes hung on the
mount)?

I'm uncomfortable using lazy unmounts as I don't have
a way of knowing if the filesystem has finally been
dropped or not.

goutham


--- Garrick Staples <garrick@usc.edu> wrote:

> On Sat, Oct 09, 2004 at 02:22:31PM -0700, Goutham
> Kurra alleged:
> > 
> > Force unmount simply isn't working on a standard
> linux
> > 2.4.22 kernel. Isn't it supposed to unmount NFS
> > filesystems no matter what?
> > 
> > In my case, the backend server is still up and
> > working, I have a "tail -f /mnt/nfs/foo" of some
> file
> > on the mount, and "umount -f" fails with "device
> or
> > resource is busy"
> 
> I don't believe force is useful in this case, what
> you want is -l.
> 
>    -f     Force unmount (in case of an unreachable
> NFS system).  (Requires
>           kernel 2.1.116 or later.)
> 
>    -l     Lazy unmount. Detach the filesystem from
> the filesystem  hierar-
>           chy now, and cleanup all references to the
> filesystem as soon as
>           it is not busy anymore.  (Requires kernel
> 2.4.11 or later.)
> 
> -- 
> Garrick Staples, Linux/HPCC Administrator
> University of Southern California
> 

> ATTACHMENT part 2 application/pgp-signature 




		
__________________________________
Do you Yahoo!?
Yahoo! Mail Address AutoComplete - You start. We finish.
http://promotions.yahoo.com/new_mail 


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: force unmount not working 2.4.22

[Michael Richardson]

-----BEGIN PGP SIGNED MESSAGE-----


>>>>> "Goutham" == Goutham Kurra <kurrag1@yahoo.com> writes:
    Goutham> I can see that '-f' tries to stop all active rpc tasks to
    Goutham> the server, but I guess the reference counts on the mount
    Goutham> are not dropped. What would be cool is a 'brute-force'
    Goutham> option that kills any process that still holds a reference
    Goutham> to the filesystem and then also unmounts it.

  A good idea, but there are still things that "kill -9" won't kill, and
people who don't understand why we need it. "Fix the application" they say.

  A better idea is that -f goes into the file table, and makes all of
the references duds. When the applications writes, they get an I/O
error, and when they read, they get EOF or I/O error.

    Goutham> The problem with using something like fuser is that it will
    Goutham> promptly hang on an unreachable server making it useless on
    Goutham> nfs filesystems.

    Goutham> Any ideas on what to do in this case (i.e, unreachable NFS
    Goutham> server, plus a bunch of processes hung on the mount)?

  Yes, a shame. Perhaps we can fix fuser as well?

    Goutham> I'm uncomfortable using lazy unmounts as I don't have a way
    Goutham> of knowing if the filesystem has finally been dropped or
    Goutham> not.

  For critical stuff, using lazy or soft mounts is not a good idea.

- --
]     "Elmo went to the wrong fundraiser" - The Simpson         |  firewalls  [
]   Michael Richardson,    Xelerance Corporation, Ottawa, ON    |net architect[
] mcr@xelerance.com      http://www.sandelman.ottawa.on.ca/mcr/ |device driver[
] panic("Just another Debian GNU/Linux using, kernel hacking, security guy"); [

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)
Comment: Finger me for keys

iQCVAwUBQWsPRIqHRg3pndX9AQH7XgP/Xx3uhOmjXZ/lVijYGSPxnldnYMo6s9Gp
0sc8IQepyaMZ9JSl/TnayvNTBjO2YXaBBliugDTqPp+RRV3lYXa0EXbTHtHRiBD0
KaGu98Z0k5iDhFrUcAcgUfU55AKDC2FtQ69YG5DhXt10EOIDtmvfcamy9ssiCRsj
No/K5Fv2IJ0=
=pzfz
-----END PGP SIGNATURE-----


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs