Hi. I want to peek inside collectd database (aka rrd files) from a remote Nagios instance. The number of systems I manage is fairly large so I am running multiple collectd "masters" which aggregate the data from various collectd clients. The theoretical rig I'm thinking is based on nrpe - on collectd masters I run nrpe server, which in turn call nagios-collectd which uses unixsock. Altough in a sandbox this works well it is a pretty long pipeline for doing monitoring. Can fail in multiple places. I can afford to run collectd and nagios on the same server. Load related.
So my options so far are: 1 extend unixsock plugin so "netsock" . if there aren't any plans in this direction i volunteer :) 2 save the rrd files on some nfs mount point and query directly from nagios 3 stick with nrpe 4 use collectd based triggers to submit passive checks to nagios 1 it would be really nice for a lot of reasons. mainly a simple pipeline and ability to query directly the collectd clients and leave the masters alone. i tried socat but it is , to me, a spartan way of supporting long running services. 2 i am too lazy to not use collectd's namespace available trough unixsock and there are some nfs-barfing-all-day aspects too 3 unreliable for my sense of job security. 4 i want to keep the alarms logic only in nagios for centralized control; but if someones has a nagios2collectd thresholds/triggers conversion script and is willing to share it would be great. i.e. i want to keep the alerting logic in nagios language not necessarly on the nagios server Thanks, _______________________________________________ collectd mailing list [email protected] http://mailman.verplant.org/listinfo/collectd
