[Info-vax] Clustering on VMS 4.7

[H Vlems]

On 18 apr, 23:59, "Robert Jarratt" <nos... at nosp.am> wrote:
> "H Vlems" <hvl... at freenet.de> wrote in message
>
> news:6e94ff26-8729-4686-a13c-fe2dab35bfd0 at z14g2000yqa.googlegroups.com...
>
>
>
>
>
> > On 16 apr, 23:42, "Robert Jarratt" <nos... at nosp.am> wrote:
> >> "H Vlems" <hvl... at freenet.de> wrote in message
>
> >>news:e6677f72-842d-48d3-a915-91420c163fe8 at f19g2000yqh.googlegroups.com...
>
> >> > On 16 apr, 21:10, "Robert Jarratt" <nos... at nosp.am> wrote:
> >> >> "H Vlems" <hvl... at freenet.de> wrote in message
>
> >> >>news:d5f99808-85f8-4722-9a19-def78a5a1b97 at o11g2000yql.googlegroups.com...
>
> >> >> > On 16 apr, 19:14, "Robert Jarratt" <nos... at nosp.am> wrote:
> >> >> >> I have set up a VMS 4.7 machine on SIMH (using the 3900 emulation)
> >> >> >> but
> >> >> >> I
> >> >> >> can't see how to set up clustering. The only cluster-related
> >> >> >> procedures I
> >> >> >> have in SYS$MANAGER are MAKEROOT.COM, BOOT_CONFIG.COM and
> >> >> >> SATELLITE_CONFIG.COM. They all tell me "This system disk is not set
> >> >> >> up
> >> >> >> as
> >> >> >> a
> >> >> >> cluster system disk." I cannot see how to get started to make the
> >> >> >> system
> >> >> >> disk a cluster system disk, can anyone tell me what the first step
> >> >> >> is?
>
> >> >> >> NB For those of you following my MicroVAX II saga, while I wait to
> >> >> >> see
> >> >> >> if
> >> >> >> I
> >> >> >> can get a replacement CPU and memory, I want to get 4.7 going on my
> >> >> >> simulated 3900 so that I am 100% sure that the DEQNA is supported
> >> >> >> when
> >> >> >> I
> >> >> >> try
> >> >> >> to net boot off it.
>
> >> >> >> Thanks
>
> >> >> >> Rob
>
> >> >> > When you installed 4.7, did the installation procedure prompt you
> >> >> > with
> >> >> > the question whether the system would be part of a cluster?
> >> >> > Later VMS versions did do that and since V5.something, even if you
> >> >> > answered NO, the system may be converted afterwards as a
> >> >> > clustermember.
> >> >> > But IIRC this was not so easy in V4.7 and earlier releases. It had
> >> >> > to
> >> >> > do with directory backlinks, I'm guessing here.... Anyway, once the
> >> >> > question was answered, you were stuck with the result or reinstall.
> >> >> > Hans
>
> >> >> No it did not ask me. I saw these prompts when installing 5.4, but I
> >> >> need
> >> >> to
> >> >> install something prior to 5.4 and the next latest I have is 4.7. I
> >> >> will
> >> >> take a look at cluster_config.com on 5.4 to see if that offers some
> >> >> clues.
>
> >> >> Regards
>
> >> >> Rob- Tekst uit oorspronkelijk bericht niet weergeven -
>
> >> >> - Tekst uit oorspronkelijk bericht weergeven -
>
> >> > I just installed 4.6 on simh and the question about a cluster member
> >> > was asked.
> >> > On your 4.7 system, do you have a SYS$SYSDEVICE:[V4COMMON] directory?
> >> > It's there on 4.6
>
> >> Now that I think about it I *was* asked and said no. This was because I
> >> was
> >> installing 4.6 with a view to updating to 4.7 and my intention had been
> >> to
> >> configure the cluster after 4.7 was installed. Is there a way to invoke
> >> that
> >> procedure again after installation? Or will I need to start again?
>
> >> Regards
>
> >> Rob- Tekst uit oorspronkelijk bericht niet weergeven -
>
> >> - Tekst uit oorspronkelijk bericht weergeven -
>
> > Rob, you've seen the difference in the names of the savesets. Even
> > numbered releases were distributed as REQUIRED. and LIBRARY. etc.
> > The odd releases, like 4.7 came as VMS047.A etc.
> > The 4.7 kit didn't really install VMS. It either patches existing
> > files or copied entirely new images from the H-kit. Sometimes these
> > new images were patched as well. The way VMS is organised on disk is
> > not affected by an update release.
> > So the answer given during the V4.6 installation is binding. Is there
> > a directory [V4CLUSTER] on your SYS$SYSDEVICE: ? (I don't think so...)
> > A system disk for a cluster was set up differently than for a stand-
> > alone node. Of course, it's all man-made, so possibly with a lot of
> > effort a non-cluster disk may be converted to a V4 system disk that
> > supports a cluster. Given the speed to install 4.6 and 4.7 on simh,
> > I'd recommend reinstalling the OS :-)
> > Hans
>
> In fact that is what I did in the end.
>
> I am curious however, back in the 80's if someone had installed VMS 4.6/7 on
> their VAX, had some users using it for some time, and then at a later date
> decided they wanted to cluster it, surely they did not have to re-install
> VMS? I have been unable to locate any clustering docs from the 4.x era
> except for this:http://www.bitsavers.org/pdf/dec/vax/vms/4.0/AA-Y512A-TE_VMS_4.0_Guid...
> which offers no clue and only applies to 4.0.
>
> I have an acquaintance who I know has some old VMS manuals I am going to ask
> him what version and see if he has a clustering manual. This is now only for
> academic interest as I have done what you suggested and reinstalled.
>
> Thanks
>
> Rob- Tekst uit oorspronkelijk bericht niet weergeven -
>
> - Tekst uit oorspronkelijk bericht weergeven -

ISTR that this topic came up during DECUS conferences (the Dutch and
the European conferences). Most sites still ran standalone VAX/VMS
systems and only very few had built cluster-capable systemdisks.
Diskspace was expensive (an RA82 was about 60.000 dutch guilders, an
RA81 40.000 IIRC and 1 guilder traded for US$ 0.40). Hence, many
system disks carried application data and user data as well. VMS
engineering told us that the conversion was supposedly possible but
the compexity was such that an automatic tool would be very difficult
to write, if not impossible. But they promised that VMS 5.0 would be
more flexible. Which it was.
VMS versions 4.7, 5.5-2 and possibly 5.4-3 were kept in use long after
newer, stable versions of VAX/VMS became available.
My own VAX systems mainly run 7.3 because UCX is mature enough to be
useful. But I dislike the V7.x output of SHOW NETWORK, need /OLD to
get the information I want.
Hans