* Paul Bijnens <paul.bijnens AT xplanation DOT com> [2005:06:17:09:51:48+0200]
scribed:
> Michael D Schleif wrote:
> >Since the tape drive change, I have wondered why I didn't seem to be
> >getting all the data on some tapes. Clearly, with Amanda, some days it
> >just doesn't want to send a full 12GB to tape; but, mostly, I have been
> >seeing <9GB going to tape, and balance sitting in holdingdisk.
> >
> >Today, I did this:
> >
> > # time sudo -u backup amtapetype -e 12g -f /dev/nst0 -o
> > Writing 32 Mbyte compresseable data: 37 sec
> > Writing 32 Mbyte uncompresseable data: 35 sec
>
> I'm pretty sure that your hardware compression is indeed off.
> Otherwise you would have a very large speed difference in writing
> uncompressed or compressed data, that is tested here. It would
> be twice or three times as fast, instead of only 2 seconds difference.
Yes, indeed. I also tried this:
# sudo /bin/mt -f /dev/nst0 compression 1
# sudo /bin/mt -f /dev/nst0 defcompression 1
# sudo -u backup amtapetype -c -f /dev/nst0 -o
Writing 1024 Mbyte compresseable data: 384 sec
Writing 1024 Mbyte uncompresseable data: 1318 sec
WARNING: Tape drive has hardware compression enabled
Estimated time to write 2 * 1024 Mbyte: 2636 sec = 0 h 43 min
> > Estimated time to write 2 * 12288 Mbyte: 26880 sec = 7 h 28 min
> > wrote 298832 32Kb blocks in 76 files in 10632 seconds (short write)
> > wrote 310628 32Kb blocks in 158 files in 10840 seconds (short write)
>
> These two lines are actually a little strange. I would have expected
> that the second pass wrote a little bit less then the first pass (and
> the difference is the space taken up by the additional filemarks).
>
> But you seem to write more in the second pass. Even 400 Mbyte.
> In previous version of amtapetype, this would be reported as a negative
> filemark size of -4603 Mbyte (the granularity of measuring is 32K).
In lieu of else to chase, how can this happen? Is this strangeness a
bad thing? How can I find the root cause?
> What actually happened is that in the first pass, there was a hard write
> error, interpreted as an end-of-tape.
> This is a symptom of an almost bad tape or tapedrive or dusty heads.
> Some OS's report an excessive soft-error rate in the kernel messages.
> Do you find anything like that in /var/log/messages?
No, there are no errors under /var/log/ for the tape drive. The tape I
am using for this was brand new, unsealed yesterday, solely for this
test. It is a hp dds-3 c5708a.
> > define tapetype unknown-tapetype {
> > comment "just produced by tapetype prog (hardware compression off)"
>
> I'm pretty sure that your hardware compression is indeed off.
> Otherwise you would have a very large speed difference in writing
> uncompressed or compressed data, that is tested here.
Yes, in recent versions of amtapetype, it actually puts it in the
comment, as above.
> > length 9522 mbytes
> > filemark 0 kbytes
> > speed 908 kps
> > }
I would be more willing to consider hardware problems on my side, if it
weren't for several extenuating circumstances:
[1] Amanda FAQ-O-Matic shows other people ending with same results; but,
I have not found documented resolution.
[2] All of my tapes in circulation have been reporting short lengths in
the `Tape Size (meg)' report field. Prior to replacing the tape
drive, these same tapes were better filled, according to these
reports.
[3] I clean the tape drive once (1x) per week. I do not notice any
unusual lights/LED's while the tape drive is working.
[4] I have successfully restored from these tapes, and using this same
tape drive, on several occasions.
Just for clarification, the brand on the tape drive is Compaq; but, the
only SDT-9000 I find on Google is the Sony brand; therefore, I assume
that Compaq is private branding the Sony drives.
What else can I do?
What do you think?
--
Best Regards,
mds
mds resource
877.596.8237
-
Dare to fix things before they break . . .
-
Our capacity for understanding is inversely proportional to how much
we think we know. The more I know, the more I know I don't know . . .
--
signature.asc
Description: Digital signature
|
