Skip to content

Conversation

@sumitagrawl
Copy link
Contributor

What changes were proposed in this pull request?

Metrics classes seperated for ICR and FCR to report seperately for prometheus

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-7464

How was this patch tested?

  1. Unit case to check report normal functionality
  2. Over prometheus, its checked for both ICR and FCR reporting

image

image

@sumitagrawl
Copy link
Contributor Author

@ChenSammi Plz review

@kerneltime
Copy link
Contributor

@neils-dev

@sumitagrawl
Copy link
Contributor Author

@neils-dev @kerneltime @ChenSammi Please review

@neils-dev
Copy link
Contributor

Thanks @sumitagrawl for filing this jira. A general comment - on the subject of this jira, Container Report at SCM is not coming separately for ICR and FCR in prometheus endpoint . Metrics for the full container report, FCR and the incremental container reports, ICR appear to be collected separately with the SCMContainerManagerMetrics. Specifically, the scm_container_manager_metrics_num_container_reports_processed_successful, scm_container_manager_metrics_num_icr_reports_processed_successful, scm_container_manager_metrics_num_container_reports_processed_failed, and
scm_container_manager_metrics_num_icr_reports_processed_failed are collected and published to the prom endpoint.

Is this jira for adding additional metrics to be collected and published for FCR and ICR ?

@sumitagrawl
Copy link
Contributor Author

Thanks @sumitagrawl for filing this jira. A general comment - on the subject of this jira, Container Report at SCM is not coming separately for ICR and FCR in prometheus endpoint . Metrics for the full container report, FCR and the incremental container reports, ICR appear to be collected separately with the SCMContainerManagerMetrics. Specifically, the scm_container_manager_metrics_num_container_reports_processed_successful, scm_container_manager_metrics_num_icr_reports_processed_successful, scm_container_manager_metrics_num_container_reports_processed_failed, and scm_container_manager_metrics_num_icr_reports_processed_failed are collected and published to the prom endpoint.

Is this jira for adding additional metrics to be collected and published for FCR and ICR ?

Its not about adding new metrics, Its about issue for reporting to prometheus. By default, prometheus endpoint takes classname as key and report metrics.
For FCR and ICR, classname is same, so causing to take one of metrics report, other is ignored. So as fix, I am separating the class for metrics for FCR & ICR to resolve this.

@ChenSammi
Copy link
Contributor

ChenSammi commented Dec 1, 2022

Hi @sumitagrawl , I have gone through the existing code. It looks like both FCR and ICR are reported in prometheus.

image

image

And both FCR and ICR event queue metrics are exposed through JMX too. I'm a little confused about this task.

@sumitagrawl
Copy link
Contributor Author

@ChenSammi
I just check, this existing metrics is ContainerManager metrics for ICR/FCR at SCM

But I have fixed for metrics related to EventQueue for ICR/FCR, this is part of FixedThreadPoolWithAffinityExecutor as earlier reported.

Both are different.

@sumitagrawl
Copy link
Contributor Author

@ChenSammi
As discussed, its a common problem for FixedThreadPoolWithAffinityExecutor and SingleThreadExecutor

Will create another PR with other implementation for this.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants