For all who are now or might in the future be running Mirror
Disc/iX:

I recently managed to get four additional hot-swap spindles for
the two HASS external disc enclosures on our 959-400.  Two
will be used to move our <long_name> from the four internal
2GB discs to two spindles in one HASS enclosure;  the other
two will be kept as hot spares.

SIDEBAR:  Empty slots in the HASS enclosure looks like a
good place to store spare spindles:  If you close the module
locking mechanism BEFORE sliding the spindle into the slot,
the locking lever comes up against the stop before any of the
pins make contact.....  Better than storing them in a cabinet
somewhere....  of course if all eight slots are full, then DNA.

Anyway, here are a couple Mirror Disc/iX factoids for today:

I wanted to cycle all four additional "new used" spindles we
just got through one "side" of our mirrored backup / test user
volume;  to verify all are good / burn them in for a few weeks;
before moving more production data over to this user volume.
Because production database logging going to this volume,
not convenient to do VSCLOSE before pulling / inserting disc
modules.

Pulled out one mirror spindle while everything was up and
running;  told system to try and build a file on it;  waited for
system to notice 74 not there;  LDEV 74 goes to DISABLED
as expected.  Go VOLUTIL and do usual REPLACEMIRRVOL.
so far all is well...  BUT:

Then tried doing same thing on LDEV 75 (while rebuild of
LDEV 74 still in process).


(1)   Pulling out the LDEV 75 spindle while REBUILD of LDEV
74 spindle is in process *appears* to start the in-process
rebuild over again..  Emphasis on "appeared";  because it was
a little hard to tell from CONSOLE messages:  One message
said ~~"DISMOUNT REQUEST REFUSED for LDEV 74"....
but then got ~~"VOLUME MOUNTED on LDEV 74"...  If any of
y'all know for sure, I would be interested in hearing....


(2)   The main "event":  After pulling out initial LDEV 75,
instead of BUILDing a new file I noticed an existing file that
could be PURGEd....  figured one should be as good as
another WRT the system "noticing" that something changed;
after which it should show LDEV 75 as DISABLED just like
LDEV 74 was.....    BZSSSST !!!!...   While I don't claim to
recall the exact sequence of my keystrokes (other than in the
case of LDEV 74 doing a BUILD while for 75 doing a PURGE),
my attempt on LDEV 75 resulted in DSTAT showing this LDEV
stuck on "BUSY".....  and note "BUSY" appears after eight
blank spaces;  after "75-" in the LDEV-TYPE column one...

About this time an error message I have never seen before
popped up on the console (failed to write down verbatim;  but
this is real close):
"VOLUME MIS-MATCH ON LDEV 75:  REMOUNT PREVIOUS
VOLUME".....
I looked though my VOLUTIL and Mirror Disc manuals;  don't
see that error message anywhere (confess not completely
shocked at failing to find it in the documentation).


Called HP RC;  after discussion thinking it might be one of a
few possibilities:

    [A]    The new hot-swap module I'm trying to bring up in
the LDEV 75 slot is bad.

    [B]    Maybe I didn't really complete PURGE of file on the
LDEV 55 - 75 mirror pair like I thought;  and system still thinks
there has been no change to LDEV 75;  even though there is a
brand new spindle in the slot (my session hung at this point (or
was on long wait);  ended up ABORTing it) ).

    [C]    There is a "weakness" or corner-case bug in Mirror
Disc software that I happened to stumble in to.

    [D]    Trying to do all above on > one volume in a mirror set
at the same time without VSCLOSE is not really supported;  i.e:
WAD.

    [E]    Something else...


O.K....  I see that REBUILD of LDEV 74 just completed...  so
now I am going back up to computer room;  and will try putting
prior 75 spindle back in;  to see what if anything that does...
then try a different "new used" hot-swap module.....  then......
well, then whatever seems appropriate when I get there...  my
basic problem remains:

How do I get LDEV 75 to switch from BUSY to DISABLED;
very preferably WITHOUT having to REBOOT....  TBD...

Interesting SIDEBAR;  and "I never knew" department:  The
RC told me that if you pull out a spindle from a running mirror
pair and there are no attempts to change a file on that spindle
after the pull, it will take 20 minutes before the spindle changes
to DISABLED in the DSTAT display....  Internal low-level timer
down in Mirror Disc software, I guess...   seems a little long to
me;  but I don't necessarily know all the factors...

Ken Sletten