Hello,
we have approximately 300 devices polled from Dude v2.2.
About 200 of them are Linux/Windows-devices, where most of them are polled via Nagios’ “check_nrpe”, which is executed with the execute-function of Dude every 30 seconds per device.
With the raising number of “check_nrpe” checks, Dude got unstable more and more. Now we have the case, that Dude crashes on every attempt to start. I think it has to do something with the great amount of simulataneously started “executions”. If i rename the “check_nrpe” in the filesystem so that Dude can’t excute it, Dude starts without crash as expected. Renaming back to it’s original name causes Dude to crash again after some time.
I have to mention that the mechanism of check-nrpe is working well (e.g. started from CommandLine).
Here the crash-log:
ACCESS VIOLATION at: 40153a
eip=40153a eflags=10246
backtrace:
stack: 3636
modules:
normis
June 5, 2008, 12:17pm
2
can you install dude v3 somewhere else and see how it works ? dude 2.2 will not be updated anymore
Hello Normis,
I used to take the “stable”-versions because of several posts claiming the bugs in the betas.
Which beta can you advice - 3b8 seems to be a bit unstable (I’ve tested it yet and have seen many errors posted by Minollie too at this thread: http://forum.mikrotik.com/t/the-dude-beta-8/20943/1 #20 )
Thanks
beta 2.2 is no more stable than any of the v3 versions. i always suggest the latest becauce I know of the issues fixed there.
Hello Normis,
I’ve migrated to the v3b8 now, seems to be stable after importing a font, but the problem with crashing while executing the many “execute”-probes still remains.
Here the crash-log:
3.0beta8 ACCESS VIOLATION at: 40153a
eip=40153a eflags=210246
log:
modules:
stack: 0 3256
Hello again,
tried to run the “execute”-probes with an agent-instance of Dude and so to reduce workload on the primary Dude.
I can run nearly every probe on the agent (is executed from it - src ip), but when switching the “execute”-probe to the agent, the poll is started from the primary Dude furthermore (src ip). I’ve checked this with ‘tcpdump’ on a linux-machine…
All of the binaries needed for the probe are available on the agent, of course.
Any idea why i can’t bring up these checks?
Thanks