Hey guys,
I recently installed Sensu and I’m wondering if it has something similar to the soft states in Nagios. Why do I ask this ? I set up a client with the check-cpu.rb plugin and it started becomming pretty spammy because it’s taking the instantaneous values of CPU usage and triggering an alert if the returned value goes above a certain threshold. The thing is that the client might be running an intensive CPU task during the check run which could stop in the next second. In that case I wouldn’t like an alert to be triggered. Any suggestions on how I can tackle this ?
Thanks,
Marius
You can send the cpu usage data to a time-series db (graphite, influxdb, opentsdb, etc) and use a check that alerts off of historical patterns in the graph. There are a couple community sensu checks that may help with this too.
···
On Mon, Feb 9, 2015 at 7:44 AM, Marius Cornea marius@remote-lab.net wrote:
Hey guys,
I recently installed Sensu and I’m wondering if it has something similar to the soft states in Nagios. Why do I ask this ? I set up a client with the check-cpu.rb plugin and it started becomming pretty spammy because it’s taking the instantaneous values of CPU usage and triggering an alert if the returned value goes above a certain threshold. The thing is that the client might be running an intensive CPU task during the check run which could stop in the next second. In that case I wouldn’t like an alert to be triggered. Any suggestions on how I can tackle this ?
Thanks,
Marius
Could you also set `refresh` if using standard handlers?
http://sensuapp.org/docs/latest/checks#common-custom-check-definitions
···
On Mon, Feb 9, 2015 at 10:07 AM, Joe Miller <joeym@joeym.net> wrote:
You can send the cpu usage data to a time-series db (graphite, influxdb,
opentsdb, etc) and use a check that alerts off of historical patterns in the
graph. There are a couple community sensu checks that may help with this
too.
On Mon, Feb 9, 2015 at 7:44 AM, Marius Cornea <marius@remote-lab.net> wrote:
Hey guys,
I recently installed Sensu and I'm wondering if it has something similar
to the soft states in Nagios. Why do I ask this ? I set up a client with the
check-cpu.rb plugin and it started becomming pretty spammy because it's
taking the instantaneous values of CPU usage and triggering an alert if the
returned value goes above a certain threshold. The thing is that the client
might be running an intensive CPU task during the check run which could stop
in the next second. In that case I wouldn't like an alert to be triggered.
Any suggestions on how I can tackle this ?
Thanks,
Marius