Hi Y'all,
We're running bacula 5.0.0 on CentOS 6. Been working just fine until
recently we saw a weird glitch.
We have a Dell ML6000 tape library (with two tape drives) that holds 39
LTO-5 tapes. We're backing up hundreds of terabytes of static data in a
series of 10TB "Full" jobs.
I loaded up all the tapes into the library then ran "label barcodes",
added them to the "Tape" pool, via drive #0, which worked great. Then
we started the backups jobs.
Each job has been completing successfully and things have been OK for
about 10 days, then finally the all the tapes in the tape library filled
up. We got the "Intervention Required" email saying no volumes were
available, that we needed to add more volumes to the "Tape" pool in
order for the current job to continue. So we unloaded all the tapes
from the library, waited for the library robot to read all the barcodes,
then waited. Nothing happened after that. So I decided I probably
needed to "label barcodes" again to add the 39 new volumes to the "Tape"
pool.
So, I ran "label barcodes" on drive 0 to the "Tape" pool. Note that
drive 0 was the drive in use for the backup jobs from before. It seemed
to load and label like 4 tapes and then all of a sudden I get this error
message saying something about "dev_in_use()" (sorry, lost the exact
message). I then noticed that bacula-sd had crashed and was no longer
running, and the job had failed.
What I *think* happened in that the "label barcodes" command had labeled
a couple tapes, then bacula tried to use one to continue the job, but
the drive was busy labeling tapes (and thus could not continue the
backup) and bacula got confused and aborted. Could that be true?
I then ran "label barcodes" again, this time choosing drive #1 when it
asked. Then began a new job, which seemed to automatically pick drive
#1 to run the job on, and not drive #0.
So I have a couple questions:
1: When I ran the "label barcodes" command on the new batch of tapes,
should I have run it from drive #1, since the job was thinking it should
use drive #0? Such that it could load the tape into drive #0 while
drive #1 could continue labeling tapes?
2: How does bacula know which drive to use to run a job? Does it
remember which drive labeled the tape, and automatically use that drive?
3: Any other way to safely label new tapes (while a job is paused)
without getting in the way of the job automatically trying to continue
once a few tapes are labeled?
Thanks for any insight!
cheers,
erich
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users
|