SYSMSH@ULKYVX.BITNET (01/11/86)
I've got a 5 node VAXcluster, one HSC50, three VAX-750's, and one 8600. Field service suggested numbering the nodes from 0 up, starting with 0 being the HSC, 1 reserved for a future HSC (hopefully), and the VAXen being 2, 3, 4, and 5. I did notice about a two minute delay when I'd take a node out of the cluster, even a little longer when quorum would be lost. I set the sysgen parameter PAMAXPORT down to 7 and did some timing tests taking things out of the cluster. When quorum is lost it doesn't seem to make much difference. It seems to make a little difference when you shut a node down, but not enough to make it significant. Go ahead and do it, make some timing tests yourself and let us know what you find. I'd like to have some kind of message come out on user terminals when quorum is lost or some other kind of cluster hangup occurs. Most of my users are from the TOPS-10 environment when a hang of any kind meant the system had crashed. They are having a difficult time getting used to a system hang that comes back in a few minutes with their process intact! Mark Hittinger Systems Programmer IV OCIS South Center University of Louisville Louisville, KY 40292 bitnet: sysmsh@ulkyvx