This seems to be holding in moderation?
* Jean-Francois Malouin <ma...@bic.mni.mcgill.ca> [20180718 08:28]:
> Hi,
>
> Suddenly, I have these pipe reply errors from one client to amanda server
> running 3.3.5. On the client amandad log file:
>
> /var/tmp/amanda/amandad/amandad.20180717220717.debug
>
> Tue Jul 17 22:07:18 2018: amandad: sending ACK pkt:
> <<<<<
> >>>>>
> Tue Jul 17 22:07:18 2018: amandad: sending PREP pkt:
> <<<<<
> OPTIONS features=ffffffff9efefbffffffffff0f;
> >>>>>
> Wed Jul 18 04:07:18 2018: amandad:
> /opt/amanda-3.3.0/libexec/amanda/sendsize timed out waiting for REP data
> Wed Jul 18 04:07:18 2018: amandad: sending NAK pkt:
> <<<<<
> ERROR timeout on reply pipe
> >>>>>
>
> I notice the exact 6 hours between sending the PREP pkt and the timeout.
> Where is this limit set?
> Serveur has dtimeout well beyond that value at 15hrs.
>
> Any hint on how to tackle this?
>
> Thanks,
> jf
Just realized that this is the same problem I experienced more than 10
years ago(!) regarding REP_TIMEOUT set to 6 hours
#define REP_TIMEOUT (6*60*60)
in amandad-src/amandad.c.
Even tough I have all the client DLEs estimates are set to server
in their dumptypes:
define dumptype "app-amgtar-span" {
"global"
program "APPLICATION"
application "app-amgtar"
priority high
allow-split
holdingdisk required
compress none
comment "tape-spanning user partitions dumped with amgtar"
# Getting bit by the REP_TIMEOUT=6hrs in amandad-src/amandad.c
# Do estimate on the server.
estimate server
}
Any way out of there?
Thanks,
jf
