Sorry for spamming. There is one additional piece of info. Notifications work well for "out-of range" situation when a client value (I experimented with cpu idle) goes in and out of range. In that case server nicely handles the notification for both "out-of-range" and "again-in-range" situations. -Vedran
On 24 April 2013 12:25, Vedran Bartonicek <[email protected]> wrote: > Hello, > I am experiencing a problem with notifications. I hope someone could help > how to fix or at least debug. > As simple setup is run with collectd (5.2.1) client and server. > This works fine: > I shut down collect client to simulate node failure. > When a value is not received from a collectd client ->notification is > writen to my exec plugin on server. > > This doensn't: > I restart collectd client, but no notification is written to the exec > plugin at server. > > > Colectd is built like: > ./configure --prefix /opt/ > --with-java=/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/ > --enable-rrdtool --enable-debug > make > > Config for thresholds: > LoadPlugin threshold > <Plugin threshold> > <Plugin memory> > <Type memory> > Instance free > FailureMin 5000000 > </Type> > </Plugin> > <Plugin disk> > Instance sda1 > <Type disk_octets> > DataSource write > WarningMax 3000 > </Type> > </Plugin> > <Plugin df> > <Type df_complex> > Instance free > WarningMin 1000000000 > </Type> > </Plugin> > </Plugin> > > > > > > > >
_______________________________________________ collectd mailing list [email protected] http://mailman.verplant.org/listinfo/collectd
