[comp.os.vms] Mounting RA60 pack volumes by searching drives

nagy%warner.hepnet@LBL.ARPA (06/15/87)

>I have two VAXes, and one HSC50 in a homogeneous cluster.  Connected to the
>HSC50 I have  RA81's and RA60's.  I would like to be able to mount the RA60's
>as  /SYSTEM/CLUSTER, but I would also link to be able to inquire about the  the
>volume name before doing it. I know how to mount/override=id, check the
>volume name with f$getdvi,  do a dismount/nounload, and then do a
>MOUNT/SYSTEM/CLUSTER.  However If the disk is already mounted with /cluster
>on one vax, when the second vax tries to mount/override,  it can't reach
>the disk.  It there a way around this?
>Any help will be greatly  appreciated.
>Tom Williams
>vtcf@ncsc.arpa

I used to do something like this on system I used manage, before those
systems became clustered.  I realized that the problem described existed
at that time.  My solution was simple; I just fiated that a particular
volume was going to be mounted ALWAYS on a particular DJAn (unit number).
This DOES NOT limit the disk pack to being mounted on a particular disk
drive since the unit plugs are easily removed and interchanged.  With
this change, all the cluster systems mount the packs as MOUNT/SYSTEM/CLUSTER
without regard to what the other systems have/have not done.  Since each
volume/unit number relationship is fixed, there is no need to "search"
for the drive with the pack and so no problem.

An aside: since I have set the mount verification timeout (MVTIMEOUT
SYSGEN parameter) to a VERY large number (10 or more hours), we have
often swapped a pack to another disk drive with the system live.  Just
take the unit offline; the systems put the volume into mount verification
and stall any I/O requests to the pack.  Spin down the pack, move it to
another drive, swap the unit plugs between the drives.  When the pack
spins back up and comes online, mount verification is cancelled and any
stalled I/O requests resume.  Works very neatly if you make a disk-to-disk
BACKUP/IMAGE of the pack and then just swap the unit plugs (very short
"downtime").

= Frank J. Nagy
= Fermilab Research Division EED/Controls
= HEPNET: WARNER::NAGY (43198::NAGY) or FNAL::NAGY (43009::NAGY)
= BitNet: NAGY@FNAL
= USnail: Fermilab POB 500 MS/220 Batavia, IL 60510