>Before the upgrade, what model tape drive were you using? I currently have
>3590-B11 and should I be investigating moving to an E J or K model
>before/during/after the upgrade.
We recently upgraded our 3494's 3590B drives to 3590E-x models, and thought
I'd share info on the experience. Note that the server Readme description
of the steps involved pretty well covers it.
We're running under ADSM server 3.1.2.50 on AIX 4.1 (and will be advancing
once our SP is fully deployed).
We boosted our Atape and atldd device driver software to the latest levels,
in advance, without having to reboot AIX. ADSM remained happy.
On the morning of the upgrade I disabled sessions and schedules in ADSM,
but left it up. I removed the drive definitions from ADSM. Then I removed
them from AIX via 'rmdev -l rmt_ -d'.
We have four 3590 drives, but only three upgrade kits had arrived intact:
the fourth was damaged by the truckers and had to be replaced...which it
turned out tooks weeks, in that this stuff has to come from Mexico. So we
proceeded to upgrade three drives and leave the fourth dormant.
The work proceeded with ADSM *and* the host system staying up: the hammerhead
design of the SCSI connections allows the SCSI chain to be disconnected but
remain intact.
It took the CEs about an hour per drive, whereafter they turned them over to
me. Did a 'cfgmgr', which didn't work well. I thought to do a Reset on the
drives, from their front panel, which resulted in a FID error message. I
had the CEs attend to that: they had neglected to properly configure the
drives. Once that was done, a 'cfgmgr' got better results - but one of the
three drives would not show up. In problem isolation, we reconnected the SCSI
to the second tail of the drive. That worked, pointing to a defective card
pack in the drive. (Blame NAFTA.) Like all other parts, it seems, this one
had to be obtained from another state - and as of this date we are still
waiting for it! We continue to operate on three drives, which is a bit tight,
but manageable.
The drive upgrades are generally working well. Like other customers, we too
are experiencing an unusual number of I/O errors. I've requested our CE to
look into it. Otherwise we are enjoying greatly increased storage capacity.
While working in the machine room, with the robot active, I noticed an unusual
"thunk" noise as the robot rolled across frames in the 3494. The young CEs,
impatient with the black doors closing slowly on the 3494 frames per their
piston damping, had been throwing their weight against the frames to force the
doors to close faster, and pushed the assembly out of alignment. Something
else to needlessly have to get fixed.
Richard Sims, BU
|
