I’m new to Grafana and I’m not sure if the proper “alerting” feature is needed. What I’m trying to do is a query that shows a table that only includes results in the table when the same nodeHostname is above the threshold (say 1) greater than or equal to 2 times.
My time series is as follows
EM_lastResponseTime{accountId="1", nodeHostname="Node1", site="siteA"} 0.160
EM_lastResponseTime{accountId="1", nodeHostname="Node1" site="siteA"} 3
EM_lastResponseTime{accountId="1", nodeHostname="Node1" site="siteA"} 2
EM_lastResponseTime{accountId="1", nodeHostname="Node1" site="siteA"} .1
EM_lastResponseTime{accountId="1", nodeHostname="Node2", site="siteA"} 0.160
EM_lastResponseTime{accountId="1", nodeHostname="Node2", site="siteA"} .2
EM_lastResponseTime{accountId="1", nodeHostname="Node2", site="siteA"} 3
EM_lastResponseTime{accountId="1", nodeHostname="Node2", site="siteA"} .1
EM_lastResponseTime{accountId="2", nodeHostname="Node3", site="siteA"} 2
EM_lastResponseTime{accountId="2", nodeHostname="Node3", site="siteA"} 3
EM_lastResponseTime{accountId="2", nodeHostname="Node3", site="siteA"} .1
EM_lastResponseTime{accountId="2", nodeHostname="Node3", site="siteA"} .4
I would want my table to include only Node1 and Node3 since in the time window I specified those have >= 2 results that are over 1. I do not want to include Node2 as the one result would be a ‘spike’.
I can’t simply do my query EM_lastResponseTime > 1 as that would include Node2. I need to do some kind of grouping operation I believe.
| accountId | time | nodeHostname | Value
| 1 | XXXXX | Node1 | 3
| 1 | XXXXX | Node1 | 2
| 2 | XXXXX | Node3 | 2
| 2 | XXXXX | Node3 | 3
I can live with the above but a bonus table would be simply showing one line for entire ACCOUNTS that have any node exceeding the threshold > X times eliminating the need to show each node. Basically I’d only show Account 1 and Account 2 in the table.
| accountId | Value
| 1 | ANY
| 2 | ANY
My data source is Prometheus.
The goal is to have a team investigate these ‘suspect’ accounts (accounts where nodes exceed a threshold more than N times). I’ve validated my data correlates w\ issues.