NFS mount attempts hangs with btrfs on server side

NFS mount attempts hangs with btrfs on server side

[Jan Engelhardt]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 695 bytes --]

Greetings.


I am observing a case whereby NFS mounting seems to hang. A tcpdump 
trace reveals that it retries forever RPC and exchanging NFS messages 
without giving any warning or error message about what's wrong. There is 
no firewall active, and the NFSserver (.1) has the client (.101) 
unconditionally allowed.

It's a simple
mount -t nfs 192.168.100.1:/B/home/jengelh /home/jengelh -o tcp,intr,soft

/B is a btrfs volume. Mounting an ext4 volume over NFS instead completes 
in a splitsecond.

Attached is the pcap file (->wireshark pretty display for deciphering).

It seems pretty much independent of the Linux kernel used; I can 
reproduce this in both 2.6.33-rc8 as well as 2.6.31.12.

[-- Attachment #2: Type: APPLICATION/X-BZIP, Size: 6644 bytes --]

Re: NFS mount attempts hangs with btrfs on server side

[manio]

 > Greetings.
 >
 > I am observing a case whereby NFS mounting seems to hang. A tcpdump
 > trace reveals that it retries forever RPC and exchanging NFS messages
 > without giving any warning or error message about what's wrong. There is
 > no firewall active, and the NFSserver (.1) has the client (.101)
 > unconditionally allowed.
 >
 > It's a simple
 > mount -t nfs 192.168.100.1:/B/home/jengelh /home/jengelh -o tcp,intr,soft
 >
 > /B is a btrfs volume. Mounting an ext4 volume over NFS instead completes
 > in a splitsecond.
 >
 > Attached is the pcap file (->wireshark pretty display for deciphering).
 >
 > It seems pretty much independent of the Linux kernel used; I can
 > reproduce this in both 2.6.33-rc8 as well as 2.6.31.12.

Hello
I have very similar behavior. On the server side i have nfs kernel server
(nfs version 3) - when mounting ext3 directories - all works fine,
but when i export btrfs directory and try to mount it:
on client side (debian sid) i've got:
mount.nfs: Connection timed out
and during mounting time on server side (debian lenny):
about 20 lines with: "authenticated mount request ..."

i also tryed it on another client with debian lenny and first mount try
was about several seconds long and results in:
mount.nfs: Stale NFS file handle
next tryes results with same error, but it was very fast responses
to the mount command

after several mount tryes i always have a problem with restarting
nfs-kernel-server with debian init script:

Stopping NFS kernel daemon: mountd nfsd.
Unexporting directories for NFS kernel daemon....
Exporting directories for NFS kernel daemon....
Starting NFS kernel daemon: nfsd <--- hungs here
... and the only way to bring the nfs server back online
was to reboot the server :(

regards,
-- 
Mariusz Bialonczyk
jabber/e-mail: manio@skyboo.net
http://manio.skyboo.net

Re: NFS mount attempts hangs with btrfs on server side

[Manio]

On 2010-04-17 22:10, manio wrote:
> > Greetings.
> >
> > I am observing a case whereby NFS mounting seems to hang. A tcpdump
> > trace reveals that it retries forever RPC and exchanging NFS messages
> > without giving any warning or error message about what's wrong. 
> There is
> > no firewall active, and the NFSserver (.1) has the client (.101)
> > unconditionally allowed.
> >
> > It's a simple
> > mount -t nfs 192.168.100.1:/B/home/jengelh /home/jengelh -o 
> tcp,intr,soft
> >
> > /B is a btrfs volume. Mounting an ext4 volume over NFS instead 
> completes
> > in a splitsecond.
>
> Hello
> I have very similar behavior. On the server side i have nfs kernel server
> (nfs version 3) - when mounting ext3 directories - all works fine,
> but when i export btrfs directory and try to mount it:
> on client side (debian sid) i've got:
> mount.nfs: Connection timed out
> and during mounting time on server side (debian lenny):
> about 20 lines with: "authenticated mount request ..."
>

Hello
Now i can reply myself - i hope it could be helpful to someone :)
The problem was: old nfs-utils and/or userland (portmap and stuff
like that). First i was trying to repeat mounting on debian sid
and it works - so i finally decided to bring up my debian stable
to squeeze - and it is working now!! :)

i am using linux-2.6.34-rc4 btw.

regards,
-- 
Mariusz Bialonczyk
jabber/e-mail: manio@skyboo.net
http://manio.skyboo.net

Re: NFS mount attempts hangs with btrfs on server side

[Jan Engelhardt]

On Wednesday 2010-04-21 12:25, Manio wrote:
> On 2010-04-17 22:10, manio wrote:
>>> Greetings.
>>>
>>> I am observing a case whereby NFS mounting seems to hang. A tcpdump
>>> trace reveals that it retries forever RPC and exchanging NFS messages
>>> without giving any warning or error message about what's wrong. There is
>>> no firewall active, and the NFSserver (.1) has the client (.101)
>>> unconditionally allowed.
>>>
>>> It's a simple
>>> mount -t nfs 192.168.100.1:/B/home/jengelh /home/jengelh -o tcp,intr,soft
>>>
>>> /B is a btrfs volume. Mounting an ext4 volume over NFS instead completes
>>> in a splitsecond.
>>
>> Hello
>> I have very similar behavior. On the server side i have nfs kernel server
>> (nfs version 3) - when mounting ext3 directories - all works fine,
>> but when i export btrfs directory and try to mount it:
>> on client side (debian sid) i've got:
>> mount.nfs: Connection timed out
>> and during mounting time on server side (debian lenny):
>> about 20 lines with: "authenticated mount request ..."
>>
>
> Hello
> Now i can reply myself - i hope it could be helpful to someone :)
> The problem was: old nfs-utils and/or userland (portmap and stuff
> like that).

I'd rather have a real description rather than "stuff like that".
Since I am not on debian, "old nfs-utils" would not happen -
rpcbind-0.1.6+git20080930 and nfs-kernel-server-1.1.3
should be very much recent enough.
Especially since exportability of filesystems is usually not
so much a userspace thing.

>First i was trying to repeat mounting on debian sid
> and it works - so i finally decided to bring up my debian stable
> to squeeze - and it is working now!! :)