| 2016-03-09 09:53:26 |
Swann Croiset |
description |
I've observed that a service which is OKAY can flap to CRITICAL sometimes. Depending on when the request is made, the "group by" returns 'null' and grafana panel interpret it as CRITICAL.
request/response:
==============
SELECT last("value") FROM "haproxy_backend_status" WHERE "hostname" =~ /node-110/ AND time > now() - 5m GROUP BY time(1m)
{"results":[{"series":[{"name":"haproxy_backend_status","columns":["time","last"],"values":[[1457193600000,1],[1457193660000,1],[1457193720000,1],[1457193780000,1],[1457193840000,1],[1457193900000,null]]}]}]}
Solution
=======
add fill(previous) to avoid null results |
I've observed that a service which is OKAY can flap to CRITICAL sometimes. Depending on when the request is made, the "group by" returns 'null' and grafana panel interpret it as CRITICAL.
request/response:
==============
SELECT last("value") FROM "haproxy_backend_status" WHERE "hostname" =~ /node-110/ AND time > now() - 5m GROUP BY time(1m)
{"results":[{"series":[{"name":"haproxy_backend_status","columns":["time","last"],"values":[[1457193600000,1],[1457193660000,1],[1457193720000,1],[1457193780000,1],[1457193840000,1],[1457193900000,null]]}]}]}
Solution
=======
add fill(previous) to avoid null results on ALL "Single stat" panel representing a service status |
|
