scott@stl.stc.co.uk (Mike Scott) (01/21/88)
I am interested in the possibility of an LAVC with 2 uVAX IIs and 2 11/750s. I am told DEC only support a 750 as boot node, which is not a suitable configuration for us. I cannot see why, apart from questions of bootstrapping the system, a 750 should not be a satellite machine. Can anyone throw any light on the problem please, and suggest how this can be done? -- Regards. Mike Scott (scott@stl.stc.co.uk <or> ...uunet!mcvax!ukc!stl!scott) phone +44-279-29531 xtn 3133.
newbery@comp.vuw.ac.nz (Michael Newbery) (01/26/88)
You can't use a 750 or 780 (or 730) as a satellite as they don't have the microcode to boot themselves off the ether. -- Michael Newbery ACSnet: newbery@vuwcomp.nz UUCP: newbery@vuwcomp Une boule qui roule tue les poules. (Landslides kill chickens)
scott@stl.stc.co.uk (Mike Scott) (02/02/88)
My thanks to all who have replied on this subject. I'm afraid that as a naive (in cluster terms) user, I asked the wrong question. The problems of boot and satellite nodes were not in fact germane, as I want each machine to have its own system disk, and to use the cluster really as a disk-sharing means. This is a rather different problem from persuading a 750 to boot from a remote disk. As some pointed out, this is easily possible as the LAVC software performs all the functions of a 'big' cluster. All you have to do is set (manually) the scssystemid, scsnode, mscp, pe3, pe6 and vaxcluster sysgen parameters and reboot. Lo and behold, the machine will cluster with any other with the same group and password! Each machine will run its own system software, and behave much as previously, except all the disks can be made available to the cluster. It seems that it is possible to cluster in this way any mix of vaxen although I have in fact only tried this on uVAX IIs as yet. By the way, does anyone know what pe3 and pe6 actually do? I set both these as boot_config.com set them, to 1. The other problems noted were that set cluster/quorum=1 (on a two-node cluster) didn't work, and during shutdown the cluster quorum was not recalculated in time to keep the sole remaining node functioning (I gather the latter problem 'will be fixed in a new release': we're on 4.6) -- Regards. Mike Scott (scott@stl.stc.co.uk <or> ...uunet!mcvax!ukc!stl!scott) phone +44-279-29531 xtn 3133.