Thanks for the responses to my problem.
Digital ended up changing everything from the SCSI controller over the
cables to the DAT itself before we got a working system! It seems like
the major problem was in the SCSI controler; it was a NCR 810, which
we changed to QLOGIC ISP1020A. Don't ask me why that made it work - my
only guess is that it is timing related.
/tonny
Some time ago I wrote:
> Hi' everybody,
>
> We use NSR 4.2B and a TLZ9L DAT autoloader from Digital as our backup
> system for our file server, but we have the extremely annoying problem
> that we very often get as little as 100Mb on a 4/8Gb DSS2 tape.
>
> We have tried a number of things (like prewriting 4Gb of data to the
> tape before use; running "mt ret" before recycling...), but nothing
> seems to help.
>
> Note the actual ammount of data on a tape varies alot: a little as
> 448kb(!!!) and as much as 4.1Gb (just once); also the same tape can
> vary a great deal (400Mb one time and 1.2Gb the next time round)...
>
> NSR writes the following messages during a backup (which all look all
> right to me):
>
> 9/13/98 23:00:01 nsrd: savegroup info: starting Backup Master (with 1
client)
> 9/13/98 23:00:22 nsrd: modi:/export/local/alpha3.2 saving to pool 'TMN'
(Backup.021)
> 9/13/98 23:06:04 nsrd: media warning: /dev/nrmt1h writing: I/O error, at
file 48 record 649
> 9/13/98 23:06:04 nsrd: media notice: 4mm 8GB tape Backup.021 on /dev/nrmt1h
is full
> 9/13/98 23:06:04 nsrd: media notice: 4mm 8GB tape Backup.021 used 228 MB of
8000 MB capacity
> 9/13/98 23:06:24 nsrd: media info: suggest relabeling Backup.022 for
writing to pool 'TMN'
>
> We never see any UERF messages that can be related to this!
>
> We have the feeling that we can write more to the tapes during
> incremental backups than during full backups, but we have no hard
> numbers to back this feeling...
>
> All this make us think, we have a hardware setup problem... But what
> is the problem exactly? Note that the autoloader is on a separate SCSI
> bus.
>
> Anybody out there with the same type of problem?
>
> The hardware configuration:
>
> AlphaServer 1000A 5/400 with
> - 512Mb memory
> - approx. 30Gb diskspace on SCSI buses
> - TLZ9L-DB on a separate SCSI bus.
>
> The (relevant?) software configuration:
>
> Digital UNIX V3.2G (Rev. 62)
> NetWorker 4.2B for Digital UNIX (BRXSOAKIT422)
Received on Tue Nov 17 1998 - 15:06:21 NZDT