Bacula backup - Autochanger
I spent the last few weeks deeply engrossed in Bacula . I set implemented a bacula backup system system for 30 Linux and Windows servers. The biggest issue was getting our new HP Storage works tape robot (Autochanger) to change the tapes once tapes became full.
I set up the whole system with clients and job definitions and specific jobs and File Sets. I was happy with the scheduling and had my tape pool and volumes settings as planned, everything seemed fine. But after two days when a 200GB tape was full bacula refused to change the tape to a new append volumes that was sitting there waiting to be used.
To cut a long story short, there where a few issues, but the main one was this.
In the “Storage” part of Bacula-dir.conf the “Device = Autochanger” must specify “Autochanger” (/dev/sg1) and not the tape drive device (/dev/nst1).
I’ll explain, a Autochanger tape robot has two devices in it. One drive that that reads and writes to the tapes, and one robotic “arm” thing that moves back and forth moving the tape from there slots to the “drive” where they can be read.
When bacula wants to change a tape in the middle of a job if a tape if full, it reads the Device = XXXXX parameter in Bacula-dir.conf to change the tape. My problem is that I had the “Drive” (/dev/nst1) device instead of “Autochanger” (/dev/sg1). And there my tape could never be changed.
Another issues worth mentioning is that some Autochangers need make bacula give them more time to actually change the tapes, so I added the “sleep 30″ to the mtx-changer script. This solved some tape change issues I had while setting up my tape pools.
vi /usr/libexec/bacula/mtx-changer
# Increase the sleep time if you have a slow device
# or remove the sleep and add the following:
wait_for_drive $device
sleep 30
exit $rtn
;;
/Roger Sinel
