Array Lost
Paul Meiners
an array meltdown, I became somewhat complacent. Other than a rare drive
failure in an array set, necessitating a drive replacement and
rebuild ( never lost an array- arrays became degraded until the dead
drive was replaced), life with the arrays has been pleasant
Two weeks ago I lost an array completely. One drive failed, I replaced
the drive; the new drive would not be recognized (on a weekend- can't
call manufacturers tech support). Re arranged the dives in the array, no
help. Finally at one point ( the drives being in the array in the
original positions) another drive in the array came up as "failed".
Ended up rebuilding the array from scratch. To top off the fun, at the
12th hour my backup software was reporting the full tapes as
differential tapes, Oh S*$% ; for 1/2 of an hour I thought somehow I
set the full backup job as a differential backup ( my fault- was not
letting the software read the full catalogue). The 12th hour is a
dangerous time- a time for mistakes and doubts. Even having the c
:partition files backed up, I still had more problems than expected.
1) Dlt tape drives are a gift to humanity
2) If your array fails completely, update your array firmware and
software drivers . I did not update the firmware in this instance, since
a call to the manufacturer ( on Monday) suggested a slight chance of
another array failure in the process
3) Always have a spare drive on hand - I always have at least one
4) Document every configuration parameter when you set up the array.
Array size, stripe size, cache setting etc. My array setting were not
lost, but if you pull your drive set to add a new set of drives, and
you have to go back to the original drives you need the setting. Also
mark your drives according to their original positions in the array
5) Always have the latest copy of your startup files.
6) A zip drive or CDR should be mandatory on a server- this was a 3
year old server with a floppy and CD.
7) Before this I , hot spare drives were not a priority ( except at my
largest clients), that has changed.
8) I found this a very positive learning experience. My disaster recover
plan needed some work. I can now have a complete meltdown and be up an
running in less than half the time of this rebuild.
Paul Meiners
Felton Green
Bummer, but thanks for the information. I've been telling folks for quite
some time to do that.
From my blurb on installation:
I also write down the types of drives, web site of all the mfr., amount of
ram, controller info, installation date, NICs and any other info that I
might need at 2:00 am when I'm brain dead. The last thing I do is use the
config reader utility and print out the server info and put it in the binder
with all the other info. This is the baseline. If I follow the above
procedure, 95% of all disk I/O errors are resolved before they can jump up
and byte me.
--
Felton Green
Novell Support Connection Volunteer SysOp
Windows NT crashed.
I am the Blue Screen of Death.
No one hears your screams.
"Paul Meiners" <pcme...@earthlink.net> wrote in message
news:39F45774...@earthlink.net...