🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: vmsnet.sysmgt
Article #25

Re: Using STARTUP$STARTUP_LAYERED

#25
From: brodie
Date: Mon, 27 Jul 1992 17:01
52 lines
3097 bytes
In article <5940@npri6.npri.com>, richard@npri6.npri.com (Richard Head) writes:
> sdroppers@pbs.org writes:
> 
>>In article <1992Jul20.135138.938@beckman.com>, dsroberts@beckman.com writes:> cluster.  He did not use the STARTUP$STARTUP_VMS and STARTUP$STARTUP_LAYERED
>>> ...stuff removed...
>>> In any case, perhaps some of you would like to share how your startup is
>>> configured so perhaps I can glean some improvements that would apply to us.
>>> 
>>We have had very good results starting the batch queues and then sending
>>jobs off in the batch queues.  We cut our reboot time from 45 minutes to about
>>15 minutes this way.  We abandoned the DEC supplied startup since we could not
>>specify the batch queue to start jobs in.
>>
>>By using batch queues we can also specify the exact order items are started.
> 
> Interestingly enough, I had the opportunity to view the startup procedures of
> a large and strategically important Digital internal VAXcluster system.
> I didn't bother to check if they were using the STARTUP$STARTUP_LAYERED data
> file scheme, but they had a *lot* of startup procedures submitted to batch
> queues.  Almost nothing was executed during SYSTARTUP_V5.  I counted at least

I use batch queues too.  (I have two single-threaded batch queues for this,
so I "sort of" have parallel startup...  i.e., any jobs that are
dependant on order (i.e., tcp/ip, mx, uucp, etc.) get submitted IN order
to a specific queue).

Another side benefit of this is kind of having a "minimum" startup
if I need to.  I.e., i may not necessarily want p1 set to 'MINIMUM',
but I certainly have the option of stopping the batch queue(s) right
away if I want to prevent further startup.

I haven't had the time to seriously investigate the SYSMAN startup
procedures (mostly because I don't see the NEED.  Our system
is a single large system.  If I was in a huge cluster, I'd probably
more seriously consider the SYSMAN stuff...)

I wonder what's next...  SYSTARTUP_v6?   :-)

> I personally don't have enough products to even bother investigating this
> bit of VMS "boot trivia", but I would favor the batch method if I did.

ditto.  (about the only thing I do DIRECTLY in SYSTARTUP is mount
my disks, define logical names, and then fire off all the other
batch jobs).   "works for me".

-- 
Kent C. Brodie - Sr. Systems Manager		InterNet:  brodie@fps.mcw.edu
Faculty Physicians & Surgeons			uucpNet:   fps.mcw.edu!brodie
Medical College of Wisconsin			MaBellNet: +1 414 266 5080

"Little bunny FOO-FOO, hoppin' through the forest,
   scoopin' up the field mice and boppin' 'em on the head.."  -childrens song
Technical Details
Message-ID: <1992Jul27.110159.674@fps.mcw.edu>
Path: arc1-ca.pugleaf.net!sparky!uunet!wupost!uwm.edu!fps.mcw.edu!brodie
References: <1992Jul20.135138.938@beckman.com> <1992Jul21.092043.16621@pbs.org> <5940@npri6.npri.com>

Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!wupost!uwm.edu!fps.mcw.edu!brodie
From: brodie@fps.mcw.edu
Newsgroups: vmsnet.sysmgt
Subject: Re: Using STARTUP$STARTUP_LAYERED
Message-ID: <1992Jul27.110159.674@fps.mcw.edu>
Date: 27 Jul 92 11:01:58 CST
References: <1992Jul20.135138.938@beckman.com> <1992Jul21.092043.16621@pbs.org> <5940@npri6.npri.com>
Distribution: world
Organization: Medical College of Wisconsin,  Milwaukee, WI
Lines: 52