[fix] [txn] Make txn metrics name conforms to the rule

[poorbarcode]

Fixes: #17921

Note:

This patch will change metrics names s_bufferedwriter_batch_record_count and s_bufferedwriter_batch_oldest_record_delay_time_second. These two names were first used in this PR #17701, and PR #17701 hasn't cherry-picked any branches yet, so this change will not cause any breaking changes.

Motivation

https://github.com/poorbarcode/pulsar/actions/runs/3156649582/jobs/5136584463
https://github.com/apache/pulsar/actions/runs/3156649597/jobs/5136596447

Problem-1

If the Prometheus-Colloctor which typed Counter is named 'xxx_count', then the output metrics-api will be named 'xxx_count_count'.

TxnLogBufferedWriterMetricsStats hits this error.

pulsar/pulsar-transaction/coordinator/src/main/java/org/apache/pulsar/transaction/coordinator/impl/TxnLogBufferedWriterMetricsStats.java

Lines 105 to 106 in fb7307d

	String recordsPerBatchMetricName =
	String.format("%s_bufferedwriter_batch_record_count", metricsPrefix);

---

Problem-2

PrometheusMetricsTest defines the standard metrics name(see code below):

["_sum", "_bucket", "_count", "_total", "_created"]

But the standard Prometheus name has three others( see: https://github.com/prometheus/client_java/blob/c28b901225e35e7c1df0eacae8b58fdfbb390162/simpleclient/src/main/java/io/prometheus/client/Collector.java#L152-L186 ):

["_info", "_gsum", "_gcount"]

pulsar/pulsar-broker/src/test/java/org/apache/pulsar/broker/stats/PrometheusMetricsTest.java

Lines 834 to 861 in fb7307d

	if (metricName.endsWith("_sum")) {
	String summaryMetricName = metricName.substring(0, metricName.indexOf("_sum"));
	if (!typeDefs.containsKey(summaryMetricName)) {
	fail("Metric " + metricName + " does not have a corresponding summary type definition");
	}
	} else if (metricName.endsWith("_count")) {
	String summaryMetricName = metricName.substring(0, metricName.indexOf("_count"));
	if (!typeDefs.containsKey(summaryMetricName)) {
	fail("Metric " + metricName + " does not have a corresponding summary type definition");
	}
	} else if (metricName.endsWith("_bucket")) {
	String summaryMetricName = metricName.substring(0, metricName.indexOf("_bucket"));
	if (!typeDefs.containsKey(summaryMetricName)) {
	fail("Metric " + metricName + " does not have a corresponding summary type definition");
	}
	} else if (metricName.endsWith("_created")) {
	String summaryMetricName = metricName.substring(0, metricName.indexOf("_created"));
	if (!typeDefs.containsKey(summaryMetricName)) {
	fail("Metric " + metricName + " does not have a corresponding summary type definition");
	}
	} else if (metricName.endsWith("_total")) {
	String summaryMetricName = metricName.substring(0, metricName.indexOf("_total"));
	if (!typeDefs.containsKey(summaryMetricName)) {
	fail("Metric " + metricName + " does not have a corresponding counter type definition");
	}
	} else {
	fail("Metric " + metricName + " does not have a type definition");
	}

---

Modifications

• Make PrometheusMetricsTest run with transaction feature
• Make txn metrics name conforms to the rule. see: https://prometheus.io/docs/practices/naming/
• Make PrometheusMetricsTest support all suffix of prometheus metrics name

Documentation

• doc-not-needed
  (Please explain why)

Matching PR in forked repository

PR in forked repository:

• [ci-run][fix][txn]Make txn metrics name conforms to the rule poorbarcode/pulsar#19

[codelipenghui]

@poorbarcode Please help confirm that we don't have a release for the metrics name that you want change.

[poorbarcode]

Hi @codelipenghui

Yes, we don't have a release for these metrics names that I want to change. Thanks

[codelipenghui]

These two names were first used in this PR #17701, and will be released in version 2.11.1

We will not release 17701 in 2.11.1 right?

[codelipenghui]

/pulsarbot run-failure-checks

[poorbarcode]

@codelipenghui

We will not release 17701 in 2.11.1 right?

Correct, PR #17701 hasn't cherry-picked any branches yet

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 47.20%. Comparing base (79a97a9) to head (55dc2c0).
⚠️ Report is 3352 commits behind head on master.

Additional details and impacted files

@@             Coverage Diff              @@
##             master   #17905      +/-   ##
============================================
- Coverage     47.24%   47.20%   -0.05%     
+ Complexity    10407    10405       -2     
============================================
  Files           692      692              
  Lines         67766    67766              
  Branches       7258     7258              
============================================
- Hits          32016    31988      -28     
- Misses        32158    32193      +35     
+ Partials       3592     3585       -7     

Flag	Coverage Δ
unittests	47.20% <ø> (-0.05%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.
see 37 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.