recently, after upgrading our bacula server from CentOS 5.10
to CentOS 6.4, we have been hit by an upstream kernel bug.
in our bacula config we use the following tape device settings
for some time.
Maximum Block Size = 1024K
Minimum Block Size = 32K
after OS upgrade the tape mount failed most of the times
16-Nov 00:03 bacula-sd JobId 431441: Error: block.c:1002 Read error on
fd=7 at file:blk 0:0 on device "Drive-1" (/dev/nst1). ERR=Device or
we got the same error using dd, but not always.
it took me about one week to find the root cause of the problem.
It is described at redhat (need a subscription to read all details)
There exists a kernel patch, but it is not released until now.
There i found also a workaround which works for us:
# cat /etc/modprobe.d/mt-st.conf
options st max_sg_segs=64
See also the latest technical notes from RHEL 6.5:
When using large block size (1MB), the tape driver sometimes
returns an EBUSY error. To work around this problem, use a
smaller block size, that is 256KB.
I wrote this just to save u time when u experiment on tape block size
>= 1024k on CentOS/RHEL 6.x.
Ulrich Leodolter <ulrich.leodolter@...>
Oesterreichische Bibliothekenverbund und Service GmbH
Raimundgasse 1/3, A-1020 Wien
Fax +43 1 4035158-30
Tel +43 1 4035158-21