More NFS troubles.

[robin_listas]

On 2012-02-19 19:26, deanjo13 wrote:
> That message will repeat on every NFS mount because of the duplicate.

It should show up in the log.

--
Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 "Celadon" at Telcontar)

[hcvv]

That's because the second attempt to mount fails and thus would not show up in the list as it only shows what has successfully mounted (the contents of mtab).

mount(8): mount filesystem - Linux man page

There is no need to explain to me what mount does. I know this allready for more then 30 years.

What I am asking you is why it shows mounts of 2000-1, 2000-2, 2000-3, 2000-5 which are not in the fstab. Until you have an explanation for it, or tell that that is also a riddle to you, I can not believe that the fstab listing and the mount listing you show have a close connection. And when they do not have a close connection, how can I draw conclusions from them?

Like Carlos I suggest that you look into dmesg output and into /var/log/messages for any information about "... keeps trying to mount the NFS shares twice" as you put it. Because looking at the window of some partitioner (as it calls itself) could point to an error in that partitioner. We should go to the bottom and use bare-bones CLI statements and logs to gather evidence.

Thus if the only symptom you have of "... keeps trying to mount the NFS shares twice" is the showing of the strange things in the partitioner, but you have no problems in the daily life of the system, may be we should look in why that partitioner shows this and not take fpr granted that it is the truth.

Also, like @Knurpht and others pointed to, it is very strange that you have to use the "_netdev" parameter at all. I can not remember any thread here with the complaint that NFS mounts are tried before the network connections are started, except of course when Networkmanager is used. I myself use NFS and the client never times out on the NFS mount at boot. Even when the server is unavailable.

[robin_listas]

On 2012-02-20 10:46, hcvv wrote:
> Also, like @Knurpht and others pointed to, it is very strange that you
> have to use the "_netdev" parameter at all. I can not remember any
> thread here with the complaint that NFS mounts are tried before the
> network connections are started, except of course when Networkmanager is
> used. I myself use NFS and the client never times out on the NFS mount
> at boot. Even when the server is unavailable.

No, I have seen that before.

There are two places where an nfs mount is activated. One is when the init
scripts read fstab and try mounting every thing there, and this happens
very early, before the network is activated; thus nfs shares will fail. In
the past, I used "noauto" for them. The _netdev keyword tell the script to
ignore those mounts.

The other place is the init script "nfs", which is dependent on network, so
it runs at the correct place mounting nfs shares.

If you look at "/etc/init.d/boot.localfs" in 11.4 (I refuse to look at 12.1
for this, I don't know systemd), you will see how the _netdev is used.

--
Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 "Celadon" at Telcontar)

[hcvv]

No, I have seen that before.

There are two places where an nfs mount is activated. One is when the init
scripts read fstab and try mounting every thing there, and this happens
very early, before the network is activated; thus nfs shares will fail. In
the past, I used "noauto" for them. The _netdev keyword tell the script to
ignore those mounts.

The other place is the init script "nfs", which is dependent on network, so
it runs at the correct place mounting nfs shares.

If you look at "/etc/init.d/boot.localfs" in 11.4 (I refuse to look at 12.1
for this, I don't know systemd), you will see how the _netdev is used.

--
Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 "Celadon" at Telcontar)

While this sounds all as a correct description, I have an NFS mount in one of my systems that has only defaults as parameter. It gives no problem either in the initial mount -a as done early in the boot, nor at the start of the nfs client later in the initialisation sequence. Never needed noaouto, nor _netdev. That is irrespective if the server is available or not.

[robin_listas]

On 2012-02-20 14:06, hcvv wrote:
> While this sounds all as a correct description, I have an NFS mount in
> one of my systems that has only -defaults- as parameter. It gives no
> problem either in the initial -mount -a- as done early in the boot, nor
> at the start of the nfs client later in the initialisation sequence.
> Never needed -noaouto, -nor -_netdev-. That is irrespective if the
> server is available or not.

Sometimes it does work, reason unknown. Other it doesn't, and the cure is
that option.

--
Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 "Celadon" at Telcontar)

[deanjo13]

.

What I am asking you is why it shows mounts of 2000-1, 2000-2, 2000-3, 2000-5 which are not in the fstab. Until you have an explanation for it.

Those I had removed at the time of the posting as I didn't want to wait 2 hours to boot to the desktop waiting for the dup mounts to mount. The relevant entries for those are identical the 2000-4 fstab line.

[twwlogin]

We have the same problem. In /etc/fstab, we have:

sanji:/home /home nfs vers=3,hard,intr,rw 0 0
sanji:/opt/dist/madara /opt/dist nfs vers=3,hard,intr,rw 0 0


After booting, we have:

$ df -k
...
sanji:/opt/dist/madara/ 726754304 0 726754304 0% /opt/dist
sanji:/home/ 802846720 76092416 726754304 10% /home
sanji:/opt/dist/madara/ 726754304 0 726754304 0% /opt/dist


This is after a fresh installation of 12.1.