netstat -i stalls indefinitely

[Victor Semaska_3]

Greetings,

I'm looking into a problem on a DS20E running V5.1A PK4. A user claims that files transferred to it via SecureShell scp from another DS20E (V5.1B PK5) are sometimes arriving corrupted. I normally don't take care of this server and the person that set it up has since left.

One of the things I tried to check is the network with the command 'netstat -i' which stalls indefinitely. Same goes for the 'netstat -an'. Here's what I see on the system:

# ps aux | grep netstat
root 302960 0.0 0.0 2.23M 216K pts/14 I + 09:40:54 0:00.00 netstat -i

What I find odd is that 10 aliases have been defined in /etc/rc.config and the same FQDN has been defined in /etc/hosts as follows:

xxx.yyy.10.1 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
#
# Aliases needed for III
#
xxx.yyy.10.2 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
xxx.yyy.10.3 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
xxx.yyy.10.4 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
xxx.yyy.10.5 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
xxx.yyy.10.6 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
xxx.yyy.10.7 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
xxx.yyy.10.8 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
xxx.yyy.10.9 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
xxx.yyy.10.10 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org
xxx.yyy.10.11 zzzzzz.vvvvvv.edu zzzzzz www.tttttt.org

Is this valid? Could this be causing the stall problem?

As for the file corruptions, I've checked all system logs at the time of the transfers and have found nothing unusual.

Thanks,
Vic

[Ivan Ferreira]

You should have a different host name for every IP address configured on the system or you may have problems (Not sure is this exact problem).

Instruct the user to use sum/cksum before and after to check the integrity of the files transfered.

[Ann Majeske]

Are you using the scp (ssh) supplied with the OS for V5.1B and the separate kit supplied by HP for V5.1A or are you using Openssh or some other vendor? There is a patch available for an scp copy problem for V5.1B:
http://www2.itrc.hp.com/service/patch/patchDetail.do?BC=patch.breadcrumb.main|patch.breadcrumb.search|&patchid=T64KIT1000096-V51BB26-E-20051110&context=tru:tru64:5.1b
V5.1A has the same problem, but since the kit is supplied separately there is not a patch available on the OS patch site. If you have a prior version support contract for the V5.1A system you can open a support call with HP to get the patch for V5.1A. If you can't get the patch, there is an easy workaround for this particular scp copy problem, if there is an existing file with the same name at the destination delete it before doing the scp.

Ann

[Ann Majeske]

Sorry, I was incorrect. V5.1A does NOT have the same scp problem as V5.1B. The scp problem I referred to is only in PK5 of V5.1B, not any other patch kit of V5.1B and not in the V5.1A kit at all. I should have checked my notes before replying :(

Ann

[Victor Semaska_3]

Ivan,

Since this is a production box I can't make changes to /etc/hosts without knowing it'll fix the problem.

I've tried duplicating the problem on a test box but no luck. So it appears using the same FQDN multiple times isn't the problem.

As for the scp problem, we're going to test the patch that Ann specified.

Thanks,
Vic