automountd not responding occasionally.

[Jeffrey L. Huff]

We have an automounted FS on an 11.i server that intermittently is not responding. I can cd into the directory 1 minute, and it is not there, try again a minute or less later and it shows up. Automounter seems to be working, just not responding "quickly" enough. How do I code for this anomily, when attempting to write data to the mountpoint that is not there?

Thanking all in advance.

[harry d brown jr]

Do you have this patch installed? PHNE_26388

live free or die
harry

[Jeffrey L. Huff]

Harry,

The patch PHNE_26388 is not applied(yet). I have it available, but it's a re-boot patch, so I need to wait a bit. But this seems to be a new problem. Neither the host server(11.i), nor the slave server have had OS changes made to them. I have attempted to look for "line quality issues", but so far have found nothing odd. Maybe it't the amount of data in the NFS mount structure?

[A. Clay Stephenson]

Keeping the patches up to date is always a good idea. You might try increasing the -t automount parameter so that filesystems are not unmounted as often. Remember, NFS is stateless, it is expected that there be delays and functions should block until the resource is available. You shouldn't have to add anything to your code to handle this situation.

If you want to actually code a method to deal with command timeouts, see this thread:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x2c4daf48a9e5d5118ff40090279cd0f9,00.html

My answer is obviously to use
a Perl wrapper for your cd or ls command. Just substitute if for my sample find command. It's easy even if you don't know Perl. There is also an example of how to do it in the shell.

Because you mention that this has just started, I would check to see if one end of one of your network connections has switched to half-duplex. Believe it or not, with a duplex or speed mismatch, network connects can continue to function, albeit badly. You might not even notice it during a telnet session but FTP or NFS transfers will be dogs.

This is easy to test. Do some fairly large (10MB or so) FTP transfers in both directions and see if the transfer rate seems reasonable.

[Jeffrey L. Huff]

Clay,

Thanks for you input. We patch servers twice a year unless it's a break/fix situation. My second pass this year is not till the end of next month. The servers in question have been up 200+ days each. I will attempt the wrapper logic, but don't know if I will be able to convince the "end user" community to try it also.

[Shannon Petry]

Some other causes in delays could be network related.

The automounter gets it's maps from NIS?

Make sure that you are on the same broadcast address, and that you can resolve hosts and ping without delays. If not make sure your /etc/resolve.conf and /etc/nsswitch.conf are correct.

Patches may fix the problem, but I usually tend to find that patches fix rare cases, not normal behavior deficiencies. Im not sure which category you fall into here.

Regards,
Shannon