Managed Switches

I have a managed switch. I am trying to mount a server using NFS. Client and server are both linux. With certain network cards the server does not mount at boot. With other network cards it does.

With the network cards that do not mount the server....if I put a sleep of 45 seconds before trying to mount at boot, it will mount successfully.

My question is: Why does the managed switch have this 45 second verification with some network cards? Basically, why does the 45 second sleep work?

Network cards that the sleep is needed are the Broadcom cards and the Intel 1000 cards.

Thanks,

 

Alright, there's quite a few variables here so I'm going to ask you a lot of questions. Are you assigning a static IP to your NFS client or a dynamic one? Is the switch acting as a DHCP server? Do you have access groups configured in the switch? Have you tried the operation over a passive switch (or your managed switch in passive mode)? If so, did you get different results? Are the client and server in the same subnet? Finally, if you don't mind could you post your server's /etc/exports?

Thanks,
-AT

 

Static IP for both the client and server.

Have not tired running in passive mode.

Exports are in the same subnet.

exports:
/home/me *(rw,no_root_squash)


The managed switches are Dell 3324 and 3348 models.

 

Unless you have some reason you want to export your home dir R/W to the whole world, I'd strongly consider changing your export to take advantage of your static IP:

Code:

/home/you 192.168.1.49(rw,root_squash)

...where '192.168.1.49' = your client's IP.

Also, try the switch in passive mode (if this is possible) just for the sake of troubleshooting. At least then you can either narrow it down to or eliminate the possibility of it being a problem with your switch's configuration.

Finally, is it possible for you to post the line in your /etc/fstab which references the NFS mount? That way we can cover all the bases.

-AT

 

Here it is:

/dev/hda2 /boot ext2 noauto,noatime 1 1
/dev/hda3 / reiserfs noatime 0 0
/dev/hda5 /mnt/local_data reiserfs noatime 0 0
/dev/hda1 none swap sw 0 0
/dev/cdroms/cdrom0 /mnt/cdrom iso9660 noauto,ro,user 0 0
/mnt/cdrom/ribstone/ribstone-packages /usr/portage/packages/All reiserfs loop,user,ro,encryption=aes-256,noauto 0 0
/dev/fd0 /mnt/floppy auto noauto,user,exec 0 0
none /proc proc defaults 0 0
none /dev/shm tmpfs defaults 0 0
192.168.27.25:/home/me /mnt/server_data nfs async,rsize=8192,wsize=8192 1 1

 

Why don't you try something along these lines and see how it works out:

Code:

192.168.27.25:/home/you /mnt/server_data nfs rw 0 0

 

What are you thinking? Just trying to limit the possible culprit?

Only Broadcom and Intel gigabit cards, have this problem. Every other NIC works fine and mounts to server w/o problems.

 

Also, if I put a sleep in netmount of 45 sec before mounting....it works.

It has to be that switch. Just not sure where the 45 delay is and why it exists.

 

I'm thinking simplify and isolate, as far as my reasoning behind the minor fstab changes are concerned. What I'm wondering is if your switches may be mangling your NFS init packets. I'm not familiar with Dell switches, as I've never had any encounters with them. But it seems that could be the problem, since different NICs give different results, but they all work with a 45-second hold-off.

 

Ya, I am not familiar with them either.

I just called Dell support and explained the problem. They stated it sounds like a configuration problem with the switch. I am taking that info for what it's worth.....I guess that just more backing for pointing the finger at the Dell switch as the problem.