Fwd: Re: [HECnet] More clustering fun

[Peter Coghlan]

I will do as you suggest. I have a disk in the box itself which is where 
the basis for the disk served by SLAVE comes from. So when I got to an 
unbootable state I have had to restore the disk.

My good friend Malcolm suggested the following which is what worked with 
me without having to autogen:

On 16/09/11 16:55, Malcolm Blunden wrote: Votes:

          Why do you run Autogen afterwards?   A change of votes doesn't 
need any other change that I can think of.   How are you changing the 
votes?   Do you understand the arcane use of CURRENT and ACTIVE by 
SYSGEN?   To set the votes to zero in a way that will apply at the next 
reboot, do

$ MCR SYSGEN
SYSGEN>   USE CURRENT
SYSGEN>   SET VOTES 0
SYSGEN>   WRITE CURRENT
SYSGEN>   ^Z

          That should fix it.   Let me know if it doesn't work!

Malcolm is absolutely right. However, you may need to do an AUTOGEN at some
point in the future and it is good to have done one that has worked before
then so that there are no surprises.

(It should have been possible to get past the joining the cluster problem
by doing a conversational boot and setting VAXCLUSTER to 0 so that the
original problem could have been found and fixed rather than having to
recreate the disk)

Regards,
Peter Coghlan.