DECServers Rebooting

[Lawrence Czlapinski]

Zeni:
1. If it were one server, I would suspect a susceptibilty to heat or a problem with the power to the terminal server. Since the reboots are simultaneous (I assume in various locations), I would update at least one of the DecServer 200's to V3.3 and see if that server stays up when the others go down. If it stays up, there is a good chance your problem is solved. The latter versions fixed crash problems. In any case, the first thing HP would say is upgrade your DecServer 200's to V3.3.
2. A bug check message would be reported on console port (port 1; 9600 baud).
" 3.3 SEVERE ERRORS
Severe errors may cause your DECserver 200 to hang or bugcheck.
Server hangs are usually recovered after 20 seconds by a automatic power fail, followed by a downline load. If this should occur, please describe as best you can the operating conditions on the server at the time of the hang.
If a FATAL BUGCHECK occurs, a bugcheck message will be printed out on the console terminal, showing the vital registers at the time of the bugcheck. Normally, an upline crash dump will automatically be created upon a fatal bugcheck error.
3. The second thing HP would say would be to upgrade to VAX VMS 7.3.
Lawrence

[Volker Halle]

Zeni,

re: Unrecognized destinations...

Ethernet packets counted as 'unrecognized destinations' will most probably be broadcast packets. The packets are being received by the DECserver (due to them being addressed to it's MAC address or an enabled multicast address or are broadcast packets). The DECserver would only expect LAT unicast or multicast packets and would not be able to handle other protocols.

Do you also see high multicast frames received ? Do you have a high percentage of broadcast messages on your LAN, maybe something like a 'broadcast storm' from time to time ?

If the DECserver would bugcheck, the bugcheck message should be indicated on the SHOW SERVER STATUS screen after the reboot (just checked the docu).

Volker.

[Dale A. Marcy]

Zeni is not in this morning, she will be in later. I checked the patches on the VAX that boots the DECServers running VMS V7.1 and it has the VAXLAT02_071 patch applied which appears to be the latest patch for V7.1 VMS LAT. Upgrading VMS is not allowed at this time because of applications having to be re-certified to run under a newer version of VMS. We are looking into upgrading the DECServer software. I am not sure how far Zeni got in getting a terminal connected to port 1 of the DECServer. The only reason Zeni mentioned the ES47s was the only change that we were aware of when the DECServers started this rebooting was a cluster reboot of 2 of the ES47s. It might have just been a fluke that it happened at the same time. They should not have any effect on the DECServers. The DECServers boot off from a VAX and are not generally used to connect to the ES47s.

[Dale A. Marcy]

Was able to capture output from Port 1 on a rebooting DECServer. I am manually typing this in, so could be subject to error:

Local -913- Fatal Bugcheck PC=184EC8, SP=05D8A8, SR=2710, MEM=000000, CODE=100
Local -905- Waiting for image dump
Local -906- Dumping to host "xx-xx-xx-xx-xx-xx"
Local -907- Image dump complete
Local -908- Resetting console terminali
Local -901- Initializing DECserver "xx-xx-xx-xx-xx-xx" -- ROM BL20, H/W Rev D.A
Local -902- Waiting for image load
Local -903- Loading from host "xx-xx-xx-xx-xx-xx"
Local -904- Image load complete

I changed the real addresses to xx and the terminali was not a typo, that is how it printed out. I have the distribution for the upgrade to V3.3, but have not been able to move it to the VAX system because of other problems (not with the VAX, but the other systems I have to use to transfer it to the VAX).

[Veli Körkkö]

Code 100 Memory parity error.

However,

MEM = The illegal memory address of an addressing error or the address of the instruction that caused the error.

nd this was 0... I would not spend time upgrading to VMS V7.3. I would much rather upgrade to latest DECserver software, i.e. install DS2033 kit to the load host(s) and reboot the terminal server.

v1.0 must be quite old since I can see V3.0 on CONDIST from Nov 1989 to March 1991, V3.1 from May 1991 to Jan 1994 and V3.3 from May 1994 to 1997 March.

_veli

[Zeni B. Schleter]

We updated the DecServer software to v3.3 and triggered a reboot of one of the servers. After 50 minutes the other two rebooted. The server with v3.3 did not reboot. This is promising.

I appreciate all the information that has be posted . It has been very helpful.

[Zeni B. Schleter]

We have not experienced any more rebooting since the DecServer 200 software was upgraded to v3.3 and loaded. Thanks for all the information and insight.