| View previous topic :: View next topic |
| Author |
Message |
Massimo B.
Veteran
Joined: 09 Feb 2005
Posts: 1568
Location: Germany
|
 Posted: Fri Jun 09, 2017 12:00 pm Post subject: NFS over autofs, lot of pending NFSv4 callback |
 |
|
Hi,
I'm mounting NFS via autofs:
| Code: | # grep nfs /etc/autofs/auto.master
/mnt/nfs /etc/autofs/auto.nfs --timeout=120 --ghost
# cat /etc/autofs/auto.nfs
data -fstype=nfs,rw,noatime,nodiratime,async,tcp,soft,fsc nfs01:/export/nfs/data
users -fstype=nfs,rw,noatime,nodiratime,async,tcp,soft,fsc nfs01:/export/nfs/users
local.SLES12 -fstype=nfs,rw,noatime,nodiratime,async,tcp,soft,fsc nfs02:/export/usr/local/run/SLES12 |
However I have a huge list of NFSv4 callback processes and increasing:
| Code: | # pgrep -alf NFS |wc -l
5196
# pgrep -alf NFS |head
310 NFSv4 callback
312 NFSv4 callback
314 NFSv4 callback
346 NFSv4 callback
347 NFSv4 callback
348 NFSv4 callback
350 NFSv4 callback
351 NFSv4 callback
352 NFSv4 callback
353 NFSv4 callback |
The number is never decreasing over uptime but getting higher. What is wrong here? The nfs mounts are working fine. I'm not even using the mounts right now. But there is one mount that always gets remounted as soon as I umount it. Umount works without issue and no log message. But 2 seconds later autofs has mounted that mount again. lsof on that mount is empty. What can I do?
P.S.: Seems to be an upstream bug: https://bugzilla.redhat.com/show_bug.cgi?id=1427493
_________________
ppc:PowerBook5,8 15"(1440)-G4/1.67,2G|amd64:HP EliteBook 8560w,i7-2620M,16G|Acer Z5610 (Core2QuadQ8200),8G|amd64-prefix:OpenSuse|HP EliteDesk 800G1 i7-4790|HP Compaq Pro 6300 i7-3770
Lila-Theme |
|
| Back to top |
|
 |
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6015
|
| Posted: Fri Jun 09, 2017 4:54 pm Post subject: |
|
|
Just checked my own system... ouch.
| Code: | ~ $ pgrep -alf NFS | wc -l
38 |
Running 4.11.3, it looks like the patch isn't in .4 either. |
|
| Back to top |
|
|
Massimo B.
Veteran
Joined: 09 Feb 2005
Posts: 1568
Location: Germany
|
| Posted: Mon Jun 12, 2017 9:23 am Post subject: |
|
|
Which patch, you mean the upstream patches from Redhat?
4.11.0-gentoo running here right now, but I've seen this issue since lot of earlier versions and also other kernels.
_________________
ppc:PowerBook5,8 15"(1440)-G4/1.67,2G|amd64:HP EliteBook 8560w,i7-2620M,16G|Acer Z5610 (Core2QuadQ8200),8G|amd64-prefix:OpenSuse|HP EliteDesk 800G1 i7-4790|HP Compaq Pro 6300 i7-3770
Lila-Theme |
|
| Back to top |
|
|
Zucca
Veteran
Joined: 14 Jun 2007
Posts: 1579
Location: KUUSANKOSKI, Finland
|
| Posted: Mon Jun 12, 2017 11:44 am Post subject: |
|
|
Whoa. This might also be the reason I've been having randomly unable to mount NFS mounts (I tend to put my system to sleep rather than shutdown, thus making several remounts on each uptime, plus all the mounts have idle timeouts). At first I thought it was the RDMA that messed things up, but I've now been mounting the shares without it and nothing seems to have changed.
_________________
..: Zucca :..
| Code: | | ERROR: '--failure' is not an option. Aborting... |
Last edited by Zucca on Fri Jun 16, 2017 8:15 am; edited 1 time in total |
|
| Back to top |
|
|
Massimo B.
Veteran
Joined: 09 Feb 2005
Posts: 1568
Location: Germany
|
| Posted: Tue Jun 13, 2017 7:08 am Post subject: |
|
|
Should we file that bug on bgo for tracking the upstream patch or even for including the patch already in gentoo-sources?
_________________
ppc:PowerBook5,8 15"(1440)-G4/1.67,2G|amd64:HP EliteBook 8560w,i7-2620M,16G|Acer Z5610 (Core2QuadQ8200),8G|amd64-prefix:OpenSuse|HP EliteDesk 800G1 i7-4790|HP Compaq Pro 6300 i7-3770
Lila-Theme |
|
| Back to top |
|
|
Zucca
Veteran
Joined: 14 Jun 2007
Posts: 1579
Location: KUUSANKOSKI, Finland
|
| Posted: Tue Jun 13, 2017 9:45 am Post subject: |
|
|
| Massimo B. wrote: | | Should we file that bug on bgo... |
There's already a bug report but it's still UNCONFIRMED. :o
_________________
..: Zucca :..
| Code: | | ERROR: '--failure' is not an option. Aborting... |
|
|
| Back to top |
|
|
peppev
n00b
Joined: 10 Aug 2009
Posts: 26
Location: Italy
|
| Posted: Tue Jun 13, 2017 11:06 am Post subject: |
|
|
| Zucca wrote: | | Massimo B. wrote: | | Should we file that bug on bgo... |
There's already a bug report but it's still UNCONFIRMED.  |
And I'm the author of this bug report:
https://bugs.gentoo.org/show_bug.cgi?id=617708
Not sure if they are the correct, definitive, patches, but my autofs gentoo systems seems
to work correctly from more than one month, now, with these patches applied to the gentoo kernel
4.9.16.
I can't understand why this bug is still in this state ;-(
G. Vitillaro. |
|
| Back to top |
|
|
Zucca
Veteran
Joined: 14 Jun 2007
Posts: 1579
Location: KUUSANKOSKI, Finland
|
| Posted: Fri Jun 16, 2017 8:17 am Post subject: |
|
|
I'm on 4.11. I wonder how long I have to wait for upstream to fix this... *sigh*
_________________
..: Zucca :..
| Code: | | ERROR: '--failure' is not an option. Aborting... |
|
|
| Back to top |
|
|
Massimo B.
Veteran
Joined: 09 Feb 2005
Posts: 1568
Location: Germany
|
| Posted: Wed Jun 21, 2017 9:50 am Post subject: |
|
|
The bug has impact on 24/7 systems with lots of NFS mounts.
A bugfix already exists.
Should we raise the priority? Is any of the upstream vanilla sources already delivering that patch?
From what I see the bug currently is requesting for information about emerge --info which doesn't make much sense for a kernel bug if not architecture specific or something.
_________________
ppc:PowerBook5,8 15"(1440)-G4/1.67,2G|amd64:HP EliteBook 8560w,i7-2620M,16G|Acer Z5610 (Core2QuadQ8200),8G|amd64-prefix:OpenSuse|HP EliteDesk 800G1 i7-4790|HP Compaq Pro 6300 i7-3770
Lila-Theme |
|
| Back to top |
|
|
Zucca
Veteran
Joined: 14 Jun 2007
Posts: 1579
Location: KUUSANKOSKI, Finland
|
| Posted: Sat Jul 01, 2017 12:22 pm Post subject: |
|
|
I wonder if this bug can be circumvented by forcing certain NFS protocol version..? Unfortnately for me, I think RDMA features can only be used with version 4 or newer. :\
_________________
..: Zucca :..
| Code: | | ERROR: '--failure' is not an option. Aborting... |
|
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Sat Jul 01, 2017 1:17 pm Post subject: |
|
|
| Massimo B. wrote: | The bug has impact on 24/7 systems with lots of NFS mounts.
A bugfix already exists.
Should we raise the priority? Is any of the upstream vanilla sources already delivering that patch? |
Gentoo definitely wants the patches, IMO; they're minimally invasive, only touching NFSv4 code, and they've been tested and reviewed by a RH bod (I think he is.)
Best not to go playing with the bug settings (leave that to the wranglers.)
If you've been hit by the bug, and the patches correct the problem (preferably tested over more than a week) then chime in on the bug and say so (briefly; don't get into any rambles, and don't keep posting. Post once to confirm the problem and the fix.)
Positive reports like that from Gentoo users will add weight to carrying the patch. (Users not enabling NFSv4 won't even notice.)
You can always vote for it, too, of course. |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Kernel & Hardware
