From bbd16c03860e9265338917c7f812ddbaa20aed58 Mon Sep 17 00:00:00 2001 From: Kern Sibbald Date: Fri, 7 Sep 2007 13:10:32 +0000 Subject: [PATCH] Update autochanger chapter --- docs/manual/autochangers.tex | 19 ++++++++++++++++++- 1 file changed, 18 insertions(+), 1 deletion(-) diff --git a/docs/manual/autochangers.tex b/docs/manual/autochangers.tex index f2dcb2c5..c5a350db 100644 --- a/docs/manual/autochangers.tex +++ b/docs/manual/autochangers.tex @@ -40,7 +40,7 @@ which is explained in more detail after this list: In version 1.37 and later, there is a new \ilink{Autochanger resource}{AutochangerRes} that permits you to group Device resources thus creating a multi-drive autochanger. If you have an autochanger, -you must use this new resource. +you {\bf must} use this new resource. Bacula uses its own {\bf mtx-changer} script to interface with a program that actually does the tape changing. Thus in principle, {\bf mtx-changer} @@ -71,6 +71,15 @@ supported by {\bf mtx} at the following link: The home page for the {\bf mtx} project can be found at: \elink{http://mtx.opensource-sw.net/}{http://mtx.opensource-sw.net/}. +Note, we have feedback from some users that there are certain +incompatibilities between the Linux kernel and mtx. For example between +kernel 2.6.18-8.1.8.el5 of CentOS and RedHat and version 1.3.10 and 1.3.11 +of mtx. This was fixed by upgrading to a version 2.6.22 kernel. + +In addition, apparently certain versions of mtx, for example, version +1.3.11 limit the number of slots to a maximum of 64. The solution was to +use version 1.3.10. + If you are having troubles, please use the {\bf auto} command in the {\bf btape} program to test the functioning of your autochanger with Bacula. When Bacula is running, please remember that for many distributions (e.g. FreeBSD, @@ -78,6 +87,14 @@ Debian, ...) the Storage daemon runs as {\bf bacula.tape} rather than {\bf root.root}, so you will need to ensure that the Storage daemon has sufficient permissions to access the autochanger. +Some users have reported that the the Storage daemon blocks under certain +circumstances in trying to mount a volume on a drive that has a different +volume loaded. As best we can determine, this is simply a matter of +waiting a bit. The drive was previously in use writing a Volume, and +sometimes the drive will remain BLOCKED for a good deal of time (up to 7 +minutes on a slow drive) waiting for the cassette to rewind and to unload +before the drive can be used with a different Volume. + \label{SCSI devices} \section{Knowing What SCSI Devices You Have} \index[general]{Have!Knowing What SCSI Devices You } -- 2.39.5