swinstall (and other SD programs) dumps core

[Erwin Mascardo]

I've got a couple of VXI chassis running 10.20. Any attempt to run almost any SD program (except swlist, which still works) produces a memory fault core dump. It worked on these machines back in August, and nothing fundamental has changed on them since then...

[James Beamish-White]

Do you have enough space in /var and /tmp for any files they may open?

Otherwise, I'd say it may need patching...

Party on dude!
James

[Sanjay_6]

Hi Erwin,

Other than checking for the patches, you can also look at the threads below to locate the source of your problem,

http://us-support.external.hp.com/cki/bin/doc.pl/sid=36fbbed20063ed5e1b/screen=ckiDisplayDocument?docId=200000024603205

http://us-support.external.hp.com/cki/bin/doc.pl/sid=36fbbed20063ed5e1b/screen=ckiDisplayDocument?docId=200000024603004

Hope this helps.

Regds

[Erwin Mascardo]

Thanks for the suggestions. No dice, though. Time to clarify a few things (at the time I posted, the bruise on my forehead was still fresh from banging it against the wall...):

* Disk space isn't a problem -- there's a couple hundred megs free on the partition where /var and /tmp reside.

* Networking is up and functioning properly. My only practical access to these machines is over the network -- I'm in DC, one machine is in Germany, and the other is in Australia...

* It does seem to be some sort of DCE/RPC error, as the core file is full of complaints about failed RPC calls. (strings is about the only resource I have for dealing with the core file, since we're talking about an HP product here.) However, like I said, the networking configuration is valid, and (again) that hasn't changed since the last time we attempted to install software in August.

I may end up patching these things after all, but I was hoping there was another solution that I was overlooking. Thanks again.

[James Beamish-White]

Have you checked swagentd is running? If not, start it from

/sbin/init.d/swagentd start

Cheers,
James

[Krishna Prasad]

I had this problem on a 10.20 system and had to re-install the sw programs....It was a few years ago and I don't have a step by step but the response center should have it. Also search the knowledge database I am almost sure you will get a hit.

[Erwin Mascardo]

Okay...while you all were getting back to me, I decided to try swgettools to reinstall the SD-UX package entirely. Still no good -- even a fresh install is dumping core on me.

Going back to patching the thing...is there a way to install a patch without using swinstall? I'm starting to get the feeling that there's an issue here that goes beyond the SD programs themselves.

[Erwin Mascardo]

Finally fixed the problem. I went and cleaned out all of the subdirectories under /var/adm/sw/queue (active/, agent/, complete/, lckfiles/, and logfiles/), and restarted swagentd. Everything worked fine after that.

My guess is that there was a leftover active session that was causing the sw programs to choke.

Thanks to everyone for their input.

[Daniel L Grahek]

If you are getting a large number of apparently DCE related errors, check to be sure the clocks on the different machines are not too far out of sync. DCE can cause a lot of problems affecting all aspects of the system if members of the cell are out of phase.

-Daniel