I have been reading this thread with interest. I too had an issue with the poller taking a fair amount of time. Granted I probably am not running near as much as some of you, but it seems if you use the poller2 instead of the old poller, things do run better.
I am posting my crontab of what I did:
------------------------------------
#Paths
JFFNMS=/opt/jffnms/engine
PHP=/usr/bin/php
# Consolidator
*/1 * * * * cd $JFFNMS && $PHP -q consolidate.php >/dev/null 2>&1
# SLAs
*/30 * * * * cd $JFFNMS && $PHP -q rrd_analizer.php >/dev/null 2>&1
# TFTP Configuration
02 4 * * * cd $JFFNMS && $PHP -q tftp_get_host_config.php >/dev/null 2>&1
# Cleanup system
02 5 * * * cd $JFFNMS && $PHP -q cleanup_raw_tables.php >/dev/null 2>&1
20 * * * * cd $JFFNMS && sh tmpwatch.sh $JFFNMS >/dev/null 2>&1
# Old/Current Poller
#*/5 * * * * cd $JFFNMS && $PHP -q poller.php >/dev/null 2>&1
# Interface Autodiscovery
#*/30 * * * * cd $JFFNMS && $PHP -q autodiscovery_interfaces.php > /dev/null 2>&1
# Experimental stuff
# Number of Poller/AD threads to run at the same time if using the new system
THREADS=5
# New Poller for big Networks
*/1 * * * * cd $JFFNMS && $PHP -q poller2.php master $THREADS >/dev/null 2>&1
# New Interface Autodiscovery (Only change is the launcher)
*/30 * * * * cd $JFFNMS && $PHP -q autodiscovery_interfaces.php master $THREADS > /dev/null 2>&1
# Network Autodiscovery
#*/30 * * * * cd $JFFNMS && $PHP -q autodiscovery_network.php master $THREADS > /dev/null 2>&1
# Stuff you will probably NOT use
# Poller Helper, Experimental
#*/1 * * * * cd $JFFNMS && $PHP -q poller_helper.php >/dev/null 2>&1
# Only in Satellite Enviorments
#* * * * * cd $JFFNMS && $PHP -q satellite_distribution.php > /dev/null 2>&1
--------------------------------
I am running jffnms 0.8.2 on a Dell GX1, Fedora 3 and polling 38 hosts currently with a total of 686 interfaces. Granted not huge, but before I started using poller2 it was a problem. Now with the machine I am using, its livable. Also, this is the only thing this machine is doing.
Doug
At 12:56 PM 4/21/2006, Jason Lavoie wrote:
On 04/21, Kory wrote:
> I ran into a similar problem but found that (in my situation) that the
> SNMP_PING function was the problem.
I noticed that started happening too, when I entered a rwcommunity for
the device. In pollers/cisco_snmp_ping_start.php:
if (($peer) && ($host_ip) && ($rw_community) && ($is_up) ...
Even with the snmp_ping disabled, the poller is too slow.
> But I do also agree that the whole process could stand to be a bit more
> efficient. I have a machine running right now that monitors SNR, RX and TX
> signal levels and traffic (~5 snmp values) on about 18,000 cable modems
> every 5 minutes using a program called Braa.
Yes, I've used and written other software that had no problem polling
large numbers of hosts. It sounds like jffnms could really benefit from
a more efficient poller, as everything else it does seems to be exactly
what we need.
-j
--
Jason Lavoie
Ratvarre sbe uver
[EMAIL PROTECTED]
-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
jffnms-users mailing list
jffnms-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/jffnms-users
