Cluster creation trouble
Last Sunday, I retried to cluster Diana and the newly arrived Aplhaserver 400, named DIDO, using CLUSTER_CONFIG_LAN procedure – the right way to do it.
It turned out there were bits and pieces wrong in VOTES and EXPECTED_VOTES system parameters – and to avoid a system hang if during the process something went wrong (due to quorum loss, Diana might be stalled) I added a quorum disk. The one tried first was the web disk, but for some reason (probably to do with voting parameters, these were not yet raised) that failed: Dido – didn’t start. So I chnaged to use the system disk, but errorenously named it “DKA100″‘ (in a way , this is correct – if another system wouldn’t have this disk). Diana came up file, buy again, Dido failed. As it turned out: Dido does have a local device “DKA100” which explains the trouble….I should have mentioned it “$116$DKA100” – the right name since it’s on the shared SCSI. (with some help from the OpenVMS ITRC). Also, I found that the SCSSYSTEMID of Diana was wrong, and had it changed to the right value.
When this all was done, I succeeded to have DIDO started from the common systemdisk – resulting in the configuration using Autogen and reboot. But there it fails – dramatically. I won’t get into detail here – just take a look on the ITRC entry. But Diana was very happily keeping up – and running.
No connection
There was a problem, though, but it wasn’t found before Monday night: NO ACCESS AT ALL to any web, and also SSH and TELNET failed: the system seemed ot of order. The kids though didn’t complain that they couldn’t access the Internet, so that seemed ok. Rushed home and checked Diana, but all seemed well: telnet, ssh, web – but locally – all was working. It was just that it seemd that all access to the EXTERNAL address was blocked: web, ssh, telnet…..Even from the local systems – or Diana itself!
Solution: reboot the router, and behold: problems done. The lesson: If anything has been done to Diana that may have an effect on exterenal access, the router needs a reset..
This means that mail between Sunday night (about 23:00) and Monday (up to about 18:00) is LOST….
Logs Ok
Scanning of the logfiles is fine now.
New licenses
The current licenses would expire 27-Jan-2007 and a new set arrived a week ago, or so. I loaded the new licenses – so I’m fit to run until 27-Jan-2008.

Published by


SYSMGR - or System Manager. On VMS, this is what a System Adminsitrator is on Unix and Windows - and a bit more. A VMS user, developer and system manager for years, alsmost as long as I'm in IT.

Leave a Reply