Eric Webb wrote:

On Sunday 10 November 2002 03:30 pm, Galen Johnson wrote:

Still having the issue with the planner. amcheck claims it can't find the
DUMP program, despite it being installed already and detected by amanda's
configure (/usr/local/sbin/dump).

amdump logs indicate problem getting results back on the estimates. I
presume this is because DUMP isn't being found, though I've tried it as
GNUTAR with no luck either.

--- log --
FAIL planner alexandria hda3 20021110 0 [missing result for hda3 in
alexandria response]
-- amdump --
error result for host alexandria disk hda3: missing estimate
--

Make sure your path is correct. If you run as root and su to amanda to
run the checks, you're only getting root's path. 'su - amanda' and
'export $PATH' to verify that you should be seeing the correct path.

I'm running this logged into user "amanda" directly. The only time root was involved was at compile/install time. User "amanda" has /usr/local/bin and /usr/local/sbin in path. What specifically is it looking for?

What do your debug logs have to say?

Another question: How do I specify device nodes for filesystems running
under LVM in linux? These are mounted from /dev/vgname/lvname... do I
just strip off the /dev and give "vgname/lvname" to Amanda?

what are they mounted to? Use the mount point itself .

Well, hang on -- I thought you were supposed to supply device nodes, not filesystem mount points. All the examples were things like "sda1" which is the device node minus the "/dev" pathing.

-E.

hmmmm....I think that the example disklist is misleading...however, if you'll look further down into it you'll see that it is looking for the _partitions_ themselves not the device nodes. I believe the Maryland folks have a strange naming scheme if this is truly what they name their partitions.

If you are indeed point to something other than the partitions then this could explain why you're timing out on the planner. Try pointing to the partition not the device node and see if it starts being happy for you.

=G=

Reply via email to