Date: Sat, 5 Feb 2000 17:16:22 -0600 (CST)
From: Don Holmgren
Subject: Re: 2.5m obslog 000204 (MJD 51579)
Comments embedded:
On Sat, 5 Feb 2000, Dan Long wrote:
> We started a bias drift (run 1110) at 00:30 APO time.  At 01:18 or so,
> we stoped the drift due to an unresponsive id1-2.
>
>
>     ID1-2 stopped archiving after about 70 frames into the first bias
> run.  A light show I've never seen before appeared on the drive.  All
> the compression lights came on ( including one which wasn't even
> labeled ), then the leds along the top started flashing in sequence
> from left to right ( note drive is mounted rotated 90 deg CW from
> horizontal ).  Below is the archReadSense from the drive.
>
>
> im> iccExec id1  archReadSense 1
> {id1 0x00000082}
> im> iccExec id1  archReadSense 2
> {id1 0x06290084}
This is a first - I've never seen such a light show.  I don't have a DLT
manual at home, but I don't believe this is covered.
>     In an attempt to clear the condition we unloaded the drive and
> power cycled it.  When it came back up the tape was reloaded and then
> the cleaning light came on.  The cleaning tape was then used, the
> light went out.  The data tape was reloaded.  All looked well except
> that the drive status remained positioning for append on ID1-2 when
> archMount was issued.  It stayed this way even when the tape stopped
> moving.  At this point an archResume was issued in the vain hope that
> things would work.  Actually ID1-1 did resume archiving.  ID1-2's
> cleaning light came on again as soon as archiving was resumed.  We
> decided to let ID1-1 finish catching u, before running the cleaning
> tape a second time in ID1-2 and put in a different tape.
>
> Feb 05 02:16:01 sdssid1    dArch      dsc_I_archTapeStat tapestat 2
> ZG41005962 JL1722    SH809M8747-215332         0 100 0 0 0 0 0 0 1129
> 20040 1 0 9994292
>
> Feb 05 02:20:24 sdssid1    dArch      dsc_I_frameArchived Frame 1110
> 76.41 was archived to tapes JL1716 (363) and (none) (4294967295)
I suspect ID1-2 is beyond hope and needs to be replaced/repaired.  It's
likely that the drive necver reported a succesful space to EOT, which we
mean the status would never change from positioning to end of tape.
>
>     We noticed that the frames in the pool are marked as archived even
> if only written to one tape.
Yes, that's correct.  The murmur log and/or tape logs must be used to
determine if frames only made it to one drive.  The archived flag only
indicates a successful write of a frame to one or both drives.
>
>     After the cleaning and tape swap ID1-2's tape would not load, or
> unload via iccExec commands.  All photo 1 tapes were unloaded and the
> crate rebooted.  When everything was back up, the tapes were
> reloaded.  ID1-2's drive came back with the light show seen earlier.
> At this point we decided to concede defeat on getting ID1-2 going.
>
>     We also couldn't repeat the process of getting ID1-1 to archive
> with ID1-2 not working.  So there are two short imager bias runs (
> 1111, 1112 ) which are not archived.  Last night's tapes are still in
> the drives pending replacement of ID1-2 or whatever is required to
> get the archiver working.  The set of 12 tapes on the table across
> from the drives are fresh tapes.
>
You should be able to use 'archNoUseDrv 2' to turn off the second drive,
and archiving should then run normally on just the first drive. This
should be done before any other commands (like archMount) since the second
drive in the described condition would likely fail all commands.
Don H.