CELESTE@BCVAX3.BITNET (08/05/87)
> We are considering using volume shadowing on our VAXcluster RA81 drives. > Please comment on your experiences and general insights if you use > volume shadowing at your site. > Specific questions: > What hardware / software versions > How many volumes shadowed, types of drives, system, non-system > Performance changes (+ or -) actually seen > Volume backup procedures for system and non-system disks > Problems I have recently installed shadowing for our common system disk(s) (2 RA82's, three node cluster) and our page and swap disk(s) (2 RA82's). We are running HSC50 v3.5 and VMS v4.5. Shadowing is version 1.0 (although if your running 4.5 you automatically have v1.+). As measured via SPM we have noted a marked improvement in our response time on the system (disks) and a marked degradation on our page and swap disks. This was to be expected according to the > 25% write rule mentioned in the shadowing manual. Our home grown nightly backup procedure had to be changed merely to add the virtual unit name rather that the physical units. Things to note: A single shadow set has to be managed by one HSC and the members should be on different requestor boards. If you are currently running VMS v4.5 you can skip the step in the installation procedure that requires you to create a new VMB.exe. You already have it. The explanation in the shadowing manual viz-a-vie standalone backup is confusing. You can shutdown your cluster and boot and run stand-alone backup in order to backup or restore your system disk. Just use the physical unit,(as configured by SAbackup), not the virtual. If you shadow your system disks you will need to be able to boot "unshadowed" for VMSupgrades. We have solved this by using an alternate boot file (the old Defboo) and one of the startup parameters startup_P8 whose value we check in systartup. VMSINSTAL runs fine with shadowing, so layered products and third party products can be installed on shadowed volumes. To shadow your system disk you need to move quorum.dat. In your systartup.com add a /NOREBUILD to the initial mount of this disk, then later on, in this same pro- cedure, do a set vol/rebuild, you'll save time booting. There is a known problem which can occur sometimes when you failover the system disk(s) to an alternate HSC. If there is I/O being done at the time you may lose a shadow set member because the shadow set is unable to recreate itself within a specified period of time. The result will be a shadow set change of state change and accompaning messages. The current fix (which works fine as this happened to me) is to merely recreate the shadow set "by hand" with the mount command. The rule here is that if you mount a disk into an existing shadow set it gets copied to, ALWAYS!!! This will get the laspsed disk and perform a full copy operation in approx. 15-20 minutes. Hope this is of helps. Celeste Finison Boston College celeste@bcvax3