What is doing the polling?
FA As far as I can tell, nothing.
I disagree. pass scripts shouldn't be run randomly. Turn on the -d
switch to figure out what's doing the polling.
Ok, I can't make sense of the log though. Excerpt below.
(disman event mib support could certainly trigger it too,
On Tue, 5 May 2009 16:04:47 +0200, Fredrik Arnerup
fredrik.arne...@edgeware.tv said:
What is doing the polling?
FA As far as I can tell, nothing.
I disagree. pass scripts shouldn't be run randomly. Turn on the -d
switch to figure out what's doing the polling.
(disman event mib support
FA Is is possible to control the interval between calls?
Using the pass_persist method is a better way to accomplish external data
than pass for exactly that reason: it leaves the process running and just
interacts with it as needed. You can even implement caching within it if
you want that as
On Tue, 5 May 2009 09:47:02 +0200, Fredrik Arnerup
fredrik.arne...@edgeware.tv said:
FA We were thinking of moving away from pass and starting using the exec method
FA instead because we don't want the sensor programs to know about OID:s, or
FA anything about snmp.
That's definitely a benefit
FA So the question remains, what triggers snmpd to call these external
FA programs?
A request to the subtree. exec calls are cached for a bit, so that will
help you there.
That's good to know.
FA When I start snmpd it seems to call them once every 4 seconds, but
FA after a while it is
On Tue, 5 May 2009 13:55:14 +0200, Fredrik Arnerup
fredrik.arne...@edgeware.tv said:
FA But the poll rate is still increasing. Now it's almost 4 calls per second
FA (66 forks per second).
What is doing the polling? Make it increase it's timeout value (use
-t 10 if it's our command line
FA But the poll rate is still increasing. Now it's almost 4 calls per
FA second
FA (66 forks per second).
What is doing the polling? Make it increase it's timeout value (use -t
10 if it's our command line tools).
As far as I can tell, nothing.
All we have is a bunch of pass statements
Wes Hardaker wrote:
The overhead you're seeing with execs is best avoidable by using a
better infrastructure. I'm sorry if you'd have to recode things, but to
be honest the speed rates are something like this (worst to best):
- pass and exec
- pass_persist
- embedded perl
- C
Not to
Hello,
We have a configuration where we have a number of sensor programs which are
called using the pass directive.
Snmpd was consuming quite a lot of cpu, so I checked the process with strace
and it seems each sensor value is being polled twice a second.
Since we have 18 sensor values, this
On Mon, 4 May 2009 11:58:48 +0200, Fredrik Arnerup
fredrik.arne...@edgeware.tv said:
FA Is is possible to control the interval between calls?
Using the pass_persist method is a better way to accomplish external
data than pass for exactly that reason: it leaves the process running
and just
10 matches
Mail list logo