DECServers Rebooting

[Zeni B. Schleter]

We have added 4 ES47s on our network. After several weeks , all of a sudden the DECServer 200s started rebooting. The reboots are simultaneous and the elapse time between reboots varies from a few minutes to an hour or more. We will be checking the network but is there anyone who has experienced the same problem? Could we have enabled something on the ES47 when turning on LAT connections that could have caused this?

The DecServers appear to reboot successfully and are functional. No errors are being reported anywhere that I can find.

Details: VAX V7.1 is doing the downloads.
New ES47s are VMS v7.3-2 . Decnet IV on
all systems.

[Robert Gezelter]

Zeni,

Obviously running a monitor on the network to capture the simulateous failure would be a good idea.

Another possible cause of the problem would be a power glitch of some kind. Any other symptoms on any other equipment in the area.

I cannot think of a reason why the presence of an ES47 would cause a problem with terminal servers. If anything caused a problem, I would suspect 7.3-2, but any such problem should have been noted a long time ago.

I hope that the above is helpful.

- Bob Gezelter, http://www.rlgsc.com

[Zeni B. Schleter]

The terminal servers are seeing a lot of Unrecognized destinations. An HP network support person said that a burst of such could exceed the terminal servers capabilities and cause them to crash. The source of such is what we will look for.

I was wondering if there were limits to the LAT services or whether the ES47 would broadcast more frequently. In setting up the LAT services we took the defaults. If a ES47 does not do anything more than any other VMS host servicing LAT, then I do not see how these hosts added to the network caused the problem we are experiencing. The ES47s replaced AlphaServers that were at the same VMS version.

I am focusing on DECNET in that we have IP traffic but I thought that should be transparent to the DECServers. That may be a wrong assumption. The DECServers are model 200. They have been in service for many many years.

Thanks for your response.

[Veli Körkkö]

what is the version of software running on those DECserver 200? I think latest would be V3.3.

I had ages ago on issue with non-DEC terminal servers after VMS upgrade. The said upgrade changed LAT from the very old "slave only LAT" to the LATmaster (master,slave capabilities) and somehow the thirdparty terminal server disliked the new LAT on VMS.

You have here though VAX/VMS V7.1 so there is not that much difference in LAT protocol on VAX V7.1 and Alpha V7.3-2 probably.

One could of course attach a terminal to the port 1 of terminal server and see what kind of messages the server issues on the "console" when dying?

_veli

[Volker Halle]

Zeni,

did you look at the SHOW SERVER STATUS display ? Escpecially the Software Status line on the bottom of the screen ? As far as I remember - and it's a long time ago - if the DECserver would crash with a bugcheck, it would record information about the bugcheck in that line.

You could compare, if all DECservers would be crashing with the same bugcheck information.

Make sure you are running the most recent available software version for the DECservers.

Do the DECserver write dumps to their load host when crashing ? If so, watch out for the disk space !

There could be a timing problem in the LAT protocol implementation on the DECserver that does not get triggered with older/slower OpenVMS systems.

Volker.

[Uwe Zessin]

Unrecognized destinations...

If I recall correctly, this means that the DECserver has received ethernet frames with protocoll types it does not understand, e.g. TCP/IP.

[Zeni B. Schleter]

DecServer versions are old v1.0

Show Status line appears to give current status and not status of crash. Did give useful info though. CPU is 6% used and memory is 12% used. It also confirmed that we are well below max number of services, nodes, and the like.

Digging up a Terminal and a Printer for Port 1 on one of the terminal servers is a good idea. The software is so old , the dumps are unreadable .

[Veli Körkkö]

well, maybe upgrading to latest (even that very old) software would be useful too.

_veli

[Robert Gezelter]

Zeni,

If the DECservers are very far behind the last version of the software issues, I would certainly upgrade at least ONE of them before going too far down the debugging path.

As I noted, the ES47 should not be doing anything special or different.

- Bob Gezelter, http://www.rlgsc.com

[Andy Bustamante]

>>> VAX VMS 7.1

There was an Alpha VMS 7.1x bug in LAT. On a "busy" network LAT packets that generated a collision and required a retransmission were padded with random values. The DECservers request retransmission of the mangled packet, the packet padded once again. . . repeat until the DecServer reboots. Depending on your network the value of "busy" may have been reached.

Check for LAT ECOs or consider an upgrade to VAX 7.3 if possible.