Skip to contentSkip to conversation

Strange issue with autofs5/fc6

DISCUSSION Michael 4 participants 7 messages Apr 18, 2007 – Apr 24, 2007 Kernel automounter

lore
Share & tools
lore

Michael

author
#0 lore raw

I am seeing some weird behavior with autofs/mount.

/proc/mounts, mounts, and auto.master don't seem to agree.

/proc/mounts:
filer:/export/users/mythtv /home/mythtv nfs rw,vers=3,rsize=262144,wsize=262144,hard,proto=tcp,timeo=600,retrans=2,sec=sys,addr=filer 0 0

mount:
filer:/export/users/mythtv on /home/mythtv type nfs (rw,addr=192.168.101.1)

Here is my auto.master
ypcat -k auto.master
/home auto.home bg,hard,intr,retry=10000,rsize=8192,wsize=8192,acregmin=3,acregmax=60,acdirmin=30,acdirmax=60,timeo=600,retrans=5,nfsvers=3,tcp

The 262144 size is strange and I don't have any idea where that is coming from.

Also, mounting by hand:

[root@client ~]# mount -o rsize=8192,wsize=8192
filer:/export/users/mythtv /mnt
[root@client ~]# mount | grep mnt
filer:/export/users/mythtv on /mnt type nfs (rw,rsize=8192,wsize=8192,addr=192.168.101.1)
[root@client ~]# grep mnt /proc/mounts
filer:/export/users/mythtv /mnt nfs rw,vers=3,rsize=262144,wsize=262144,hard,proto=tcp,timeo=600,retrans=2,sec=sys,addr=filer 0 0

Any thoughts on what is happening here? Is rsize really 262144 or are things being reported incorrectly?

[root@client ~]# uname -a
Linux client 2.6.20-1.2944.fc6 #1 SMP Tue Apr 10 17:46:00 EDT 2007 x86_64 x86_64 x86_64 GNU/Linux
[root@client ~]# rpm -q autofs
autofs-5.0.1-0.rc3.26

Michael

maintainer
#1 lore raw
Quoted 32 lines · Michael

I am seeing some weird behavior with autofs/mount.

/proc/mounts, mounts, and auto.master don't seem to agree.

/proc/mounts:
filer:/export/users/mythtv /home/mythtv nfs rw,vers=3,rsize=262144,wsize=262144,hard,proto=tcp,timeo=600,retrans=2,sec=sys,addr=filer 0 0

mount:
filer:/export/users/mythtv on /home/mythtv type nfs (rw,addr=192.168.101.1)

Here is my auto.master
ypcat -k auto.master
/home auto.home bg,hard,intr,retry=10000,rsize=8192,wsize=8192,acregmin=3,acregmax=60,acdirmin=30,acdirmax=60,timeo=600,retrans=5,nfsvers=3,tcp

The 262144 size is strange and I don't have any idea where that is coming from.

Also, mounting by hand:

[root@client ~]# mount -o rsize=8192,wsize=8192
filer:/export/users/mythtv /mnt
[root@client ~]# mount | grep mnt
filer:/export/users/mythtv on /mnt type nfs (rw,rsize=8192,wsize=8192,addr=192.168.101.1)
[root@client ~]# grep mnt /proc/mounts
filer:/export/users/mythtv /mnt nfs rw,vers=3,rsize=262144,wsize=262144,hard,proto=tcp,timeo=600,retrans=2,sec=sys,addr=filer 0 0

Quoted 2 lines

Any thoughts on what is happening here? Is rsize really 262144 or are things being reported incorrectly?

I don't know.
I haven't noticed this before but that's certainly odd. Maybe the folks on nfs-list will be able to help.

Ian

#2 lore raw
In reply to Michael

==> On Wed, 18 Apr 2007 09:21:45 -0700, Michael <[email protected]> said:

michael> I am seeing some weird behavior with
michael> autofs/mount. /proc/mounts, mounts, and auto.master don't michael> seem to agree.

michael> /proc/mounts: filer:/export/users/mythtv /home/mythtv nfs
michael> rw,vers=3,rsize=262144,wsize=262144,hard,proto=tcp,timeo=600,retrans=2,sec=sys,addr=filer michael> 0 0

michael> mount: filer:/export/users/mythtv on /home/mythtv type nfs michael> (rw,addr=192.168.101.1)

michael> Here is my auto.master ypcat -k auto.master /home auto.home
michael> bg,hard,intr,retry=10000,rsize=8192,wsize=8192,acregmin=3,acregmax=60,acdirmin=30,acdirmax=60,timeo=600,retrans=5,nfsvers=3,tcp

michael> The 262144 size is strange and I don't have any idea where michael> that is coming from.

Can you send along an automount debug log? Perhaps automount is not passing the [rw]size parameters to mount. For information on gathering debug logs, see:
http://people.redhat.com/jmoyer/

-Jeff

author
#3 lore raw
Quoted 10 lines · Jeff Moyer

==> On Wed, 18 Apr 2007 09:21:45 -0700, Michael <[email protected]> said:

michael> The 262144 size is strange and I don't have any idea where michael> that is coming from.

Can you send along an automount debug log? Perhaps automount is not passing the [rw]size parameters to mount. For information on gathering debug logs, see:
http://people.redhat.com/jmoyer/

I will go capture the logs. However, it does appear to be a kernel/mount issue.

If I do the mount by hand and not using autofs it still has the same problem.

root@client /]# mount -o rsize=16384,wsize=16384,proto=tcp,timeo=600,retrans=2 filer:/export /mnt

mount | grep mnt
filer:/export on /mnt type nfs (rw,rsize=16384,wsize=16384,proto=tcp,timeo=600,retrans=2,addr=192.168.101.1)

grep mnt /proc/mounts
filer:/export /mnt nfs rw,vers=3,rsize=262144,wsize=262144,hard,proto=tcp,timeo=600,retrans=2,sec=sys,addr=filer 0 0

So I don't know if it is just reporting incorrectly or if it is actually mounted incorrectly.

I will also open an issue on bugzilla for fc6.

Michael

#4 lore raw

==> Regarding Re: [autofs] Strange issue with autofs5/fc6; Michael <[email protected]> adds:

Quoted 3 lines · michael> Jeff Moyer wrote:
Quoted 3 lines

==> On Wed, 18 Apr 2007 09:21:45 -0700, Michael <[email protected]> said:

michael> The 262144 size is strange and I don't have any idea where that is michael> coming from.

Quoted 4 lines
Quoted 4 lines

Can you send along an automount debug log? Perhaps automount is not passing the [rw]size parameters to mount. For information on gathering debug logs, see: http://people.redhat.com/jmoyer/

michael> I will go capture the logs. However, it does appear to be a michael> kernel/mount issue.

michael> If I do the mount by hand and not using autofs it still has the michael> same problem.

michael> root@client /]# mount -o
michael> rsize=16384,wsize=16384,proto=tcp,timeo=600,retrans=2 michael> filer:/export /mnt

michael> mount | grep mnt filer:/export on /mnt type nfs
michael> (rw,rsize=16384,wsize=16384,proto=tcp,timeo=600,retrans=2,addr=192.168.101.1)

michael> grep mnt /proc/mounts filer:/export /mnt nfs
michael> rw,vers=3,rsize=262144,wsize=262144,hard,proto=tcp,timeo=600,retrans=2,sec=sys,addr=filer michael> 0 0

michael> So I don't know if it is just reporting incorrectly or if it is michael> actually mounted incorrectly.

Oh, I somehow missed that.

michael> I will also open an issue on bugzilla for fc6.

OK, it's probably best to use component kernel and assign it to [email protected].

-Jeff

#5 lore raw
In reply to Michael
Quoted 4 lines · Michael

root@client /]# mount -o rsize=16384,wsize=16384,proto=tcp,timeo=600,retrans=2 filer:/export /mnt

Quoted 8 lines

mount | grep mnt
filer:/export on /mnt type nfs (rw,rsize=16384,wsize=16384,proto=tcp,timeo=600,retrans=2,addr=192.168.101.1)

grep mnt /proc/mounts
filer:/export /mnt nfs rw,vers=3,rsize=262144,wsize=262144,hard,proto=tcp,timeo=600,retrans=2,sec=sys,addr=filer 0 0

Is the same volume from the same server already mounted somewhere else in this box? If so, the kernel is sharing the superblock for these two mounts of the same volume, and both the mount flags and nfs parameters have been set by the first mount.

For more information, check upstream commit
54ceac4515986030c2502960be620198dd8fe25b.

Cheers,
Fábio

Signature · 1 line
ex sed lex awk yacc, e pluribus unix, amem
author
#6 lore raw
Quoted 10 lines · Fabio Olive Leite

Is the same volume from the same server already mounted somewhere else in this box? If so, the kernel is sharing the superblock for these two mounts of the same volume, and both the mount flags and nfs parameters have been set by the first mount.

For more information, check upstream commit
54ceac4515986030c2502960be620198dd8fe25b.

Yes, the volume is mounted multiple times on the machine. However, they all report bogus values.

How am I supposed to know what values are being used if I can't trust /proc/mounts for any mounts?

Michael