HP3000-L Archives

September 1998, Week 1

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Jeff Kell <[log in to unmask]>
Reply To:
Jeff Kell <[log in to unmask]>
Date:
Tue, 1 Sep 1998 19:47:45 -0400
Content-Type:
text/plain
Parts/Attachments:
text/plain (67 lines)
I was working on bringing our new 969 to life today, and after balancing
out the remaining disk configs, building nmconfig from scratch, and
other little tasks.  DNS was up, telnet was up, inetd was rolling,
services configured, NS/3000 working, 100TX card working at 100Mbps and
full duplex as hoped, now let's configure volume sets.

We have two internal 4Gb Seagates, only one of which was configured in
the <long-name> volume set.  A Jamaica array was populated with 8x4Gb
Seagates of a slightly different flavor.  First step, extend the system
volume set.  Bearing in mind I've been burned by the old fragmentation
issues with 2-spindle system volume sets, I intended to limit the
permanent space on the new member.  So:

volutil: newvol mpexl_system_volume_set:member2 2 75 25
[The are you sure [y/n]? message... Y

DISMOUNT REQUEST GRANTED FOR LDEV2 (or whatever, I'm going by memory)

UNKNOWN VOLUME MOUNTED ON LDEV2
*Error:  Subsys 146(?); Info=-6   (I can get more detailed tomorrow)

So now, SHOWSET <long-name> VOLUMES says 2 members but :DSTAT shows
ldev2 to be UNKNOWN.  OK, doesn't volutil: initvol fix this case?
Nope, another cryptic error.  OK, formatvol 2.  This works (but takes
about 90 minutes).  initvol again fails.  Call HPRC.  Suggestion -
reinstall from System Preload Tape.  Aaaargh!  Get sysgen >tape with
@[log in to unmask]@ store appended to save where I'm at.

OK, preload tape installs, but no NMCONFIG and I/O config doesn't match.
Work through getting I/O configs back, restore NMCONFIG from my tape,
finally SYSMAP and ODE> MAPPER are happy with the drives.  OK, before
doing the Preload Tape restore, let's expand the system volume set:

volutil: newvol mpexl_system_volume_set:member2 2 75 25

(Poof!  Same error again!)  Call back to HPRC and get another engineer
who was much more knowledgeable.  To get to the point...

"newvol" is rather sensitive with respect to the system volume set.
When processing the newvol, if things "look" OK, it increments the
member count in the VSIT, but then "something" fails and it doesn't back
out the VSIT change.  You're dead in the water with the system volume
set, it's INSTALL time again (or hire Stan for some lengthy
bit-twiddling of your VSIT :-) as described in one ESC SR I found).

The problem is particular to Seagate drives (according to her) when you
place limits on perm/trans allocation.  She suggested reinstalling from
*my* tape (which was already close to functional) and doing the newvol
again specifying only the member and ldev.  Sure enough, it worked fine.

To confuse matters, I was able to create (at least) a user volume master
on one of the Jamaica drives with a similar 90/10 split of perm/trans
(yes, user volumes never have trans, but I like a cushion for perm space
shortages).  Didn't try newvol to see if it suffered the same fate as
ldev 2 (but could be easily cleaned up, so I might tomorrow).

They are different models of Seagate drives (internal vs Jamaica).  The
system still isn't quite healthy, on my sysgen SLT [log in to unmask]@.@ store, I
forgetfully omitted ;DIRECTORY.  I've been hacking at this thing and
other daily duties for 14 hours today and I think it will have to wait
until tomorrow to recover the directory structure :-)

Beware of volutil newvol, particularly with Seagates!  I've done more
installs and reboots today than the past 10 years (well, almost).

Jeff Kell <[log in to unmask]>

ATOM RSS1 RSS2