question Re: Cloudbreak Autoscaling doesn't check all metric alarms in Support Questions

Cloudbreak Autoscaling doesn't check all metric alarms

mdovhan — Thu, 23 Jan 2020 16:24:12 GMT

I created 12 metrics alarms and policy configurations based on these alarms (6 for scale out and 6 for scale in). But in cloudbreak logs I can see that only several alarms are checked and then "No scaling activity required".
Why doesn't cloudbreak check all metrics alarms? What the number of checked alarms are depend on? How can I change this behavior?

Re: Cloudbreak Autoscaling doesn't check all metric alarms

mdovhan — Fri, 24 Jan 2020 10:17:54 GMT

The current Policy configuration for Autoscaling:

However in cloudbreak logs I can see:

[cb-stack-id:1] Checking metric based alert: 'memory-used-critical' [cb-stack-id:1] Finished cronTimeEvaluator for cluster 1 in 72 ms [cb-stack-id:1] Ambari 'alertHistory' finished in 79 ms [cb-stack-id:1] Checking metric based alert: 'memory-used-ok' [cb-stack-id:1] Ambari 'alertHistory' finished in 58 ms [cb-stack-id:1] Alert: memory-used-ok is in 'OK' state since 1027.78 min(s) [cb-stack-id:1] Ambari 'clusterHosts' finished in 350 ms [cb-stack-id:1] No scaling activity required [cb-stack-id:1] Finished metricEvaluator for cluster 1 in 581 ms

Cloudbreak checks only "memory-used-critical" and "memory-used-ok" but ignores everything else.
Cloudbreak version is 2.9.1.

Re: Cloudbreak Autoscaling doesn't check all metric alarms

mdovhan — Fri, 24 Jan 2020 14:05:51 GMT

The problem has been fixed in the next way:
- create all policies for scaling out first

- then create all policies for scaling in

After checking first policy for scaling in (memory-used-ok) I can see "No scaling activity required" message in the log but now cloudbreak checks all my "critical" policies.