Hi Arno, I've been doing some tests as well as updating the firmware for both SCSI card and the Autoloader... just in case. >> > Thanks for replying. I've actually tested the mtx-changer script (by >> > default at /etc/bacula/mtx-changer) and it works fine. I can list the >> > number of slots, load and unload specific slots fine. > Have you tried accessing the tape *immediately* after the load command > returns? If I issue the mtx-changer commands manually there's no problem. My drive takes about 45 from loading a tape to reading the Volume. I already have a bacula instance running on a 2nd identical SureStore418 autoloader. But its running Bacula 1.36.3 which works fine. >> > However I've found >> > that even though the documentation mentions you can use 2 parameters, >> > somehow the trap does not work. So in essence mtx-changer /dev/sg2 >> > unload will not work. Instead you have to use mtx-changer /dev/sg2 >> > unload 1 /dev/nst0 0 where (1 is where the tape was loaded from). > I don't know why this is the case - here it always worked with two > parameters only - but it's quite a while that I used an unmodified > version of mtx-changer, so this might have changed I've resorted to manually doing it because the btape autoloader commands have failed. However it has to be a configuration problem somewhere as it works fine when I connect to my production backup server running bacula 1.36.3
>> > I've even checked by issuing the correct mtx commands that the >> > mtx-changer script does and that works too. However running the btape >> > test (btape -c /etc/bacula-sd.conf /dev/nst0) and then running the >> > autochanger command will fail. It just keeps on loading tape 1, >> reading >> > for 10 minutes and then unloading, then reloading, reading ... repeat. >> > Each time getting longer. I've even labeled the tape using the console >> > label command first to ensure it wasn't an empty tape and its still >> fails... >> > >> > If I run it from the console (not btape) then I get the following >> > emailed as an error:- >> > >> > 03-Jul 13:44 FC5-test: Error: bnet.c:426 Write error sending 55 >> bytes to client:127.0.0.1:36131: ERR=Broken pipe >> > >> > Don't know why the pipe is breaking... The passwords are all >> correct. I've even checked that there's no firewall in the way (not >> supposed to matter because its localhost). > I don't, too, but that might follow from the underlying problem with > tape access... From my current tests I suspect when bacula issues a mount, it's wrongly interpreting that there's something in the drive when there's nothing loaded. Normally when I run status storage I should get a BLOCKED status if the drive is not loaded. But now I get drive loaded with 0. I'm looking into the mtx-changer script (last part) with the loaded command. Something has to be missing there that's somehow not working with my setup. I've changed the timeout to 60 seconds (45 should work but 15 more for some padding...) I've used both rpm installation and compile from source right now. Both exhibit similar problems. >> > Is there anything else I should be looking at? > Try running the SD with debug output, and watch for the autochanger > operations. > I've enabled to mtx.log and it seems to be issuing the correct commands. But somehow the mount command fails. > Arno >> > Thanks! >> > >> > Regards, >> > Anwar Hmmm.... you mentioned earlier about 15 seconds timout? Is this a compile variable or the timeout in the mtx-changer script? Thanks! Kind Regards, Anwar ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
