| Title: | VAX and Alpha VMS |
| Notice: | This is a new VMSnotes, please read note 2.1 |
| Moderator: | VAXAXP::BERNARDO |
| Created: | Wed Jan 22 1997 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 703 |
| Total number of notes: | 3722 |
These days it's often impossible to avoid creating a multi system disk cluster - for example when mixing architectures. For most cluster environment files (SYSUAF, RIGHTSLIST LMF$LICENSE etc...), it's quite easy to move them to a disk other than the system disk, but I cannot find a way to share a common VMS$LAYERED.DAT file across multiple system disks. SYSMAN assumes that the file is common, so you can't just SET ENV/CLU to have STARTUP commands distributed to other nodes. The file is also opened before any user modifiable files are executed, so it's not possible to redirect the file by logical names. I prefer to manage clusters using SYSMAN, but it doesn't seem possible if there's more than one system disk. Any tricks I've missed? John Gillings, Sydney CSC
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 621.1 | AUSS::GARSON | DECcharity Program Office | Wed May 21 1997 03:03 | 18 | |
re .0
You can apparently redefine STARTUP$STARTUP_LAYERED as part of the VMS
startup and STARTUP.COM will explicitly close the file that it opened
for the layered startup and open the redefined file. Unless there is
stuff in the layered startup that executes prior to DEVICES then
defining this logical name in SYLOGICALS.COM should work reliably. If
you require redefinition earlier I suppose you could add a step to the
VMS INITIAL phase i.e. your own command procedure but you would have to
establish that the disk (with the common layered startup) was actually
accessable at that point.
If this gets you a common VMS$LAYERED.DAT then presumably SYSMAN will
respect this logical name.
NB: I didn't test any of this so caveat emptor.
HTH
| |||||
| 621.2 | Thanks, that will fix it | GIDDAY::GILLINGS | a crucible of informative mistakes | Wed May 21 1997 22:07 | 16 |
Thanks Derek,
Kind of a weird concept, check that the file is there, then allow
the user to change the file, then check if it's been changed, close
and reopen. Why not just wait and see if there when it's really
needed? Avoid several possible error states!
It means I'll have to leave a dummy, file on each system disk to avoid
an error message.
I think I'll define the logical name in SYLOGICALS.COM, but also
copy the "master" version back to the local SYS$STARTUP. That way I
should have a nearly-always-up-to-date version on each system disk
in case the system is booted standalone.
John Gillings, Sydney CSC
| |||||
| 621.3 | AUSS::GARSON | DECcharity Program Office | Thu May 22 1997 20:05 | 16 | |
re .2
It's not really that wierd. What seems to be happening is that the
startup will execute the INITIAL phase for VMS and then the INITIAL
phase for LAYERED and then the DEVICES phase for VMS and then the
DEVICES phase for LAYERED etc. Any one of those phase+file (file being
VMS or LAYERED) occurrences can redefine STARTUP$STARTUP_LAYERED and
the startup procedure wants that change to apply to any subsequent
phase(s). In practice it is not likely to make much difference because
layered startup has no contribution to the early phases but in theory
the code is sound.
If you are going to copy the master back to the local SYS$STARTUP you
might want to output a warning message when the local copy is used on
some particular startup that the layered startup is possibly slightly out
of date.
| |||||