Are OpenSearch's t3.small.search EC2 instances using the unlimited or standard (CPU credits-based) configuration?

Question

I've been experiencing an issue where my OpenSearch Service managed instances (t3.small.search type) periodically experience downtime of around 30 minutes. Since my node's CPU usage is on average higher than the baseline but my memory pressure appears to be a non-issue, I was wondering if these instances might be running out of CPU credits and being throttled into an unresponsive state. Are OpenSearch's managed t3 EC2 instances running in unlimited mode? And if CPU credits are involved with these instances, shouldn't OpenSearch be making the CPUCreditBalance and similar metrics available (as it has done in the past for t2 instances)?

Answer

Hello dsi-krosecrans,

by default t3 instances launch in unlimited mode.

https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/burstable-performance-instances-unlimited-mode.html

`T4g, T3a and T3 instances launch as unlimited by default (unless you change the default).`

I couldn't find Information specifically for OpenSearch and also no specific metrics like you wrote, so  I guess that unlimited mode counts for t3.search-instances too.

I found a metric but that is only available for t2-instances:

`CPUCreditBalance - This metric is available only for the T2 instance types.`

https://docs.aws.amazon.com/opensearch-service/latest/developerguide/managedomains-cloudwatchmetrics.html

Sincerely
Heiko

Are OpenSearch's t3.small.search EC2 instances using the unlimited or standard (CPU credits-based) configuration?

Relevant content