Re: Handitarded....odd (partial) estimate timeout errors.
2006-01-04 19:49:46
--On January 4, 2006 7:20:50 PM -0500 Jon LaBadie <jon AT jgcomp DOT com> wrote:
Asking questions about things I know nothing ... :)
Are you using iptables?
If so, have you installed and configured the ??conntrack?? module?
Paul asked for the logs, it seems like there's an amanda bug. The units in
question are attached to the same broadcast domain/VLAN and are in the same
subnet, so are talking directly to eachother. It's not an obvious network
or switch problem going on. I thought maybe an MTU limit of 1500 bytes but
apparently amanda is set to fragment UDP packets up to 64k and so that
should be fine, and other drives are making it. Anyway thanks anyway Jon
:) I think we've hit some sort of bug or something in amandad, or planner
(I think it sends the SERVICE sendsize packets) or both.
Network wise BTW the backup server is connected to a switch here in the
office, which is trunked further to a switch upstairs, then to another
switch in the blade chassis, then to the untrunked connection to the
(amanda backup client) nfs server which is the one having issues. It
seemed maybe some sort of odd packet size limit or some other 'max number
of' limit in planner, since planner is sending duplicate requests sorta for
the affected DLEs.
Anyway....I'm getting a headache thinking about it :) all my other DLEs
seem ok for that host, and the ones that it misses are not always exactly
the same, but all seem to be non-calcsize estimated.
|
|
|