Hi,
24.06.2008 18:37, RA Cohen wrote:
>
> Arno Lehmann wrote:
>> Hi,
>>
>> 22.06.2008 21:15, RA Cohen wrote:
>>
>>> Here is a current status snapshot of my problem installation: [It is now
>>> Sunday afternoon June 22, a full backup was scheduled Friday, with
>>> differential for Saturday and Sunday]
>>>
>>> Bacula seems to be waiting for a signal of some kind from the drive.
>>> Does the output of sta sd look normal? I assume the maximum jobs
>>> exceeded message is there because the tape did not recycle. It is in
>>> fact defined for a max of 9 jobs, 3 jobs for each day.
>>>
>>>
>> ...
>>
>>> *sta sd=InternalQuantum
>>> Connecting to Storage daemon InternalQuantum at 10.150.8.240:9103
>>>
>>> fserv-sd Version: 2.2.8 (26 January 2008) i386-portbld-freebsd6.2
>>> freebsd 6.2-RELEASE-p8
>>> Daemon started 20-Jun-08 16:04, 0 Jobs run since started.
>>> Heap: heap=212,992 smbytes=87,479 max_bytes=87,991 bufs=106 max_bufs=108
>>> Sizes: boffset_t=8 size_t=4 int32_t=4 int64_t=8
>>>
>>> Running Jobs:
>>> Writing: Full Backup job fservNightlySave JobId=1899 Volume="Friday-0002"
>>> pool="FridayPool" device="InternalQuantum" (/dev/sa0)
>>> spooling=0 despooling=0 despool_wait=0
>>> Files=0 Bytes=0 Bytes/sec=0
>>> FDSocket closed
>>> ====
>>>
>>> Jobs waiting to reserve a drive:
>>> 3610 JobId=1899 Volume max jobs exceeded on drive "InternalQuantum"
>>> (/dev/sa0).
>>> ====
>>>
>>> Terminated Jobs:
>>> JobId Level Files Bytes Status Finished Name
>>> ===================================================================
>>> 1886 Full 1 171.7 M OK 18-Jun-08 07:04 BackupCatalog
>>> 1887 Full 0 0 Cancel 19-Jun-08 08:47
>>> fservNightlySave
>>> 1890 Full 0 0 Error 19-Jun-08 08:55
>>> fservNightlySave
>>> 1891 Full 493 350.5 M Error 19-Jun-08 08:56
>>> mservNightlySave
>>> 1893 Full 184,332 76.88 G OK 19-Jun-08 10:23
>>> fservNightlySave
>>> 1894 Full 10,069 3.326 G OK 19-Jun-08 10:30
>>> mservNightlySave
>>> 1895 Full 1 173.3 M OK 19-Jun-08 10:30 BackupCatalog
>>> 1896 Full 184,399 76.92 G OK 20-Jun-08 09:50
>>> fservNightlySave
>>> 1897 Full 10,061 3.333 G OK 20-Jun-08 09:57
>>> mservNightlySave
>>> 1898 Full 1 174.5 M OK 20-Jun-08 09:58 BackupCatalog
>>> ====
>>>
>>> Device status:
>>> Device "FileStorage" (/tmp) is not open.
>>> Device "InternalQuantum" (/dev/sa0) is mounted with:
>>> Volume: Friday-0002
>>> Pool: *unknown*
>>> Media type: SDLT
>>> Total Bytes Read=129,024 Blocks Read=2 Bytes/block=64,512
>>> Positioned at File=0 Block=0
>>> ====
>>>
>>> In Use Volume status:
>>> Friday-0002 on device "InternalQuantum" (/dev/sa0)
>>> Reader=0 writers=0 reserved=0 released=1
>>> ====
>>>
>>>
>> The above looks all normal (like the SD working correctly), though I
>> would have expected that Bacula asks for a useable tape in this situation.
>>
>> I don't know if you expect the loaded tape to get recycled; if you do
>> and it doesn't recycle, check with 'llist volume=Friday-0002' if it
>> should be recycleable.
>>
>> If that is not the problem, you have to find why Bacula doesn't use
>> the tape drive; I find the combination of the drive being mounted with
>> a volume and, at the same time, being released a bit astonishing.
>>
>> What happens if you issue a mount command?
>>
>> Arno
>>
>>
> Arno,
>
> What happens when mount is issued: it came back with the "waiting on
> Storage" error.
Ok. Admittedly, not really ok, but at least that seems to confirm that
what you see is really what Bacula thinks the state of the volumes
should be.
> I cannot show this to you now as I have had to try other
> things. The drive itself has been replaced with one that has Quantum's
> latest and greatest firmware - guess what - same result.
Well, I'm not sure this has to do with the hardware. Rather it looks
like Bacula itself deadlocks with the maximum volume jobs setting.
> So I am looking
> at the Volume max jobs because of the message 'Volume max jobs
> exceeded.' It seems bacula is looking first at that Max Jobs value
> before proceeding to recycle the tape (even though the catalog was
> marked for the tape/volume to be recycled) The example I presented in
> this thread included three full backup jobs on Friday, three
> differential on Saturday, and three on Sunday, a total of 9 jobs. Two of
> the jobs are typical, the third is the catalog backup. It doesn't matter
> whether it's the Friday tape, or the Monday or Tuesday etc. I always ran
> this situation [with my Quantum 40/80 and no problems] with Max Vol Jobs
> = 9 on Friday (really Friday thru Sunday 1 tape), and Max Vol Jobs = 3
> Monday thru Thursday (4 tapes). When I set it to 2, the tape recycles
> properly, but then I'm left waiting to mount a new tape for the third
> job. The only reason for the Max Vol Jobs was to get the tape to become
> Used. So now I am trying no value for Max Vol Jobs, and setting Volume
> Use Duration for 23h to see if this solves what I believe is still
> basically a hardware <-> bacula issue.
I'd like to see more details about the SD-tape drive interaction...
what I recall from your mails is not related to drive problems, I
believe. Still I understand that with your DLT drive you don't observe
this behaviour.
>
> These Quantum 160/320 SDLT are known to be tricky and finicky as
> compared to the 40/80 DLT, unfortunately, but I have to find a way to
> live with them as I have 4 of them!
Send them here ;-)
> I'll report on my success or failure - I appreciate any and all support.
> It is sometimes hard to describe concisely just what is going on...
Yup.
Although I'm a bit short with time others might have some useful
ideas, too, so I'd recommend to capture a debug trace of the SD when
it gets into this state. Perhaps in the debug output something shows up...
Arno
--
Arno Lehmann
IT-Service Lehmann
www.its-lehmann.de
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users
|