If the Pandora FMS installation goes down from time to time, or some remote checks (network, plugins, WMI, Web monitors) sometimes return incorrect values (for example, an ICMP check should give 1
and returns 0
) or you obtain a remote check that inexplicably remains in unknown mode, it may be due to the following causes:
- Insufficient Random Access Memory (RAM). To use Pandora FMS on a production system, it should have at least 2 GB. It can be used on systems with less memory, but you have to “tune” all the components finely, and even so, you’ll be limited to a very small environment, with very few threads / sub networks and a MySQL that consumes very little memory.
- Threads importance. In Pandora FMS we can configure the amount of threads of each one of the PFMS servers, if for example, in the Pandora FMS servers view in the Web Console, we observe that some of our servers have a queue, we can increase the amount of threads for a better data processing. They are modified from the configuration file
/etc/pandora/pandora_server.conf
:snmpconsole_threads
network_threads
plugin_threads
wmi_threads
recon_threads
dataserver_threads
web_threads
inventory_threads
export_threads
icmp_threads
snmp_threads
prediction_threads
syslog_threads
provisioningserver_threads
alertserver_threads
ncmserver_threads
- Please refer to the official documentation for all current parameters.
- Remember to restart the PFMS server after making changes to its configuration file for them to be effective:
/etc/init.d/pandora_server restart
.
- Virtual environments. In some misconfigured virtual environments there exist instability problems that occur when a there’s a lot of load on the real physical system. This can be seen on the system log (dmesg) as
pandora_server coredumps
or through MySQL.
- See also: