Fix Request_Handler_Avg_Idle_Percent metric bug #12289
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #12289 +/- ##
=========================================
Coverage 74.97% 74.98%
- Complexity 6640 6642 +2
=========================================
Files 373 373
Lines 25355 25355
Branches 3404 3404
=========================================
+ Hits 19010 19012 +2
Misses 4953 4953
+ Partials 1392 1390 -2 🚀 New features to boost your workflow:
|
| { | ||
| "datasource": "${DS_PROMETHEUS}", | ||
| "expr": "sum(kafka_server_kafkarequesthandlerpool_requesthandleravgidlepercent_total{namespace=\"$kubernetes_namespace\",strimzi_io_cluster=\"$strimzi_cluster_name\",kubernetes_pod_name=~\"$strimzi_cluster_name-$kafka_broker\"}*100) by (kubernetes_pod_name)", | ||
| "expr": "sum by (kubernetes_pod_name) (rate(kafka_server_kafkarequesthandlerpool_requesthandleravgidlepercent_total{namespace=\"$kubernetes_namespace\",strimzi_io_cluster=\"$strimzi_cluster_name\",kubernetes_pod_name=~\"$strimzi_cluster_name-$kafka_broker\"}[5m]) / 1e9 * 100)", |
There was a problem hiding this comment.
In order to make the query responsive to how users interact with the dashboard, I would replace the fixed [5m] with [$__rate_interval]. I was also wondering if irate (instantaneous rate) would be better in this case, as we want to know the current state of the nodes.
There was a problem hiding this comment.
Thanks @fvaleri. Will try irate and investigate results.
There was a problem hiding this comment.
The use of irate vs rate is a genuine question from me to people with more experience with PromQL. To me it looks like a better fit for this kind of metric.
cc @mimaison
There was a problem hiding this comment.
Here is an image of the graph using irate instead of rate if that helps:
There was a problem hiding this comment.
Whichever function to prefer kind of depends. I think a more important aspect for the sample dashboards is consistency. It seems irate() is use for most graphs in this dashboard, so I'd stick to that for this query.
There was a problem hiding this comment.
Agreed that irate() is more consistant with the rest of the queries.
There was a problem hiding this comment.
@OwenCorrigan76 fyi, the [$__rate_interval] part of my comment is not yet addressed.
OwenCorrigan76
force-pushed
the
Fix_RequestHandlerAvgIdlePercent_metric_bug
branch
from
c36b79c to
56a2667
Compare
Signed-off-by: OwenCorrigan76 <owencorrigan76@gmail.com>
Signed-off-by: OwenCorrigan76 <owencorrigan76@gmail.com>
OwenCorrigan76
force-pushed
the
Fix_RequestHandlerAvgIdlePercent_metric_bug
branch
from
56a2667 to
06d1049
Compare
Signed-off-by: OwenCorrigan76 <owencorrigan76@gmail.com>
OwenCorrigan76
force-pushed
the
Fix_RequestHandlerAvgIdlePercent_metric_bug
branch
from
06d1049 to
9624045
Compare
|
@ppatierno This does not have the required approvals!!! |
I got confused, my bad (no need for 3 exclamation marks though ;-)) If any @strimzi/maintainers doesn't agree with the change please open another PR to revert it back or propose a different solution. |
5 participants
Type of change
Description
This PR addresses the follwoing issue:
#12126
The
Request Handler Avg Idle Percentdashboard is now showing a percentage average instead of an ever increasing counter.Checklist
Please go through this checklist and make sure all applicable tasks have been done