TEZ-4371 Implement ClientServiceDelegate.getJobCounters

[lewismc]

After a number of years, I decided to revisit TEZ-4371.
When the Tez client is used as the MapReduce client (via YARNRunner), ClientProtocol.getJobCounters(JobID) was returning empty Counters. This change implements counter retrieval so Tez DAG Counters are exposed as MapReduce Counters, unblocking NUTCH-2839 (e.g. Nutch Injector job counters on Tez).
I added some unit tests to this PR. My local testing continues with Nutch master branch (1.23-SNAPSHOT)

[tez-yetus]

💔 -1 overall

Vote	Subsystem	Runtime	Logfile	Comment
+0 🆗	reexec	5m 5s		Docker mode activated.
			_ Prechecks _
+1 💚	dupname	0m 0s		No case conflicting files found.
+0 🆗	detsecrets	0m 0s		detect-secrets was not available.
+1 💚	@author	0m 0s		The patch does not contain any @author tags.
+1 💚	test4tests	0m 0s		The patch appears to include 1 new or modified test files.
			_ master Compile Tests _
+1 💚	mvninstall	8m 40s		master passed
+1 💚	compile	0m 23s		master passed
+1 💚	checkstyle	1m 6s		master passed
+1 💚	javadoc	0m 30s		master passed
+0 🆗	spotbugs	0m 58s		tez-mapreduce in master has 124 extant spotbugs warnings.
			_ Patch Compile Tests _
+1 💚	mvninstall	0m 13s		the patch passed
+1 💚	codespell	0m 25s		No new issues.
+1 💚	compile	0m 13s		the patch passed
+1 💚	javac	0m 13s		the patch passed
+1 💚	blanks	0m 0s		The patch has no blanks issues.
+1 💚	checkstyle	0m 8s		the patch passed
+1 💚	javadoc	0m 11s		the patch passed
+1 💚	spotbugs	0m 34s		the patch passed
			_ Other Tests _
-1 ❌	unit	1m 3s	/patch-unit-tez-mapreduce.txt	tez-mapreduce in the patch passed.
+1 💚	asflicense	0m 12s		The patch does not generate ASF License warnings.
		20m 31s

Reason	Tests
Failed junit tests	tez.mapreduce.client.TestClientServiceDelegate

Subsystem	Report/Notes
Docker	ClientAPI=1.54 ServerAPI=1.54 base: https://ci-hadoop.apache.org/job/tez-multibranch/job/PR-464/1/artifact/out/Dockerfile
GITHUB PR	#464
Optional Tests	dupname asflicense javac javadoc unit spotbugs checkstyle codespell detsecrets compile
uname	Linux e529f4865a5b 5.15.0-141-generic #151-Ubuntu SMP Sun May 18 21:35:19 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux
Build tool	maven
Personality	/home/jenkins/jenkins-home/workspace/tez-multibranch_PR-464/src/.yetus/personality.sh
git revision	master / 093177b
Default Java	Ubuntu-21.0.10+7-Ubuntu-124.04
Test Results	https://ci-hadoop.apache.org/job/tez-multibranch/job/PR-464/1/testReport/
Max. process+thread count	239 (vs. ulimit of 5500)
modules	C: tez-mapreduce U: tez-mapreduce
Console output	https://ci-hadoop.apache.org/job/tez-multibranch/job/PR-464/1/console
versions	git=2.43.0 maven=3.8.7 spotbugs=4.9.3 codespell=2.4.1
Powered by	Apache Yetus 0.15.1 https://yetus.apache.org

This message was automatically generated.

[abstractdog]

yay! happy to see you working on this, let me find some time to review this next week

[abstractdog]

I was worried that items are not evicted from dagClientMap, but something similar is happening in case of ClientCache too:

tez/tez-mapreduce/src/main/java/org/apache/tez/mapreduce/client/ClientCache.java

Line 40 in 9725e38

//TODO: evict from the cache on some threshold

maybe we can take care of this in a followup ticket

[abstractdog]

@lewismc : this patch looks good and neat! I'm wondering if you can show any proof of manually testing this? I'm just asking, I remember back in 2020 when NUTCH-2839 was created, I somehow discovered the empty counters problem with an MR job submitted by YARNRunner, I wish I could do the same experiment easily now (of course, haven't made notes back then :D ), so if you already experimented with this, that would help a lot with approving and merging this, thanks in advance!

[lewismc]

Hi @abstractdog thanks for taking a look. RE: testing, yes that is ongoing and I will follow-up with an end-to-end example which exercises individual Nutch MapReduce jobs running as dual-vertex Tez DAG's.

My test environment (I'll link the docker compose project soon) utilizes the official Apache Hadoop 3.4.3 Docker image which is both interesting and challenging for a few reasons.

1. Hadoop 3.4.3 cluster Java 11 runtime
2. Tez 1.0.0-SNAPSHOT Java 21 compile-time dependency (with Java 11 runtime flag), this literally required reintroeucing some Java 11 logic into the codebase... I will go into more detail with a follow-up patch
3. Nutch 1.23-SNAPHOT which was also compiled with Java 21 with Java 11 runtime flag.

This is somewhat messy but it will be cleared up soon when Hadoop 3.5.0 is released and we can upgrade in both Tez and Nutch and drop Java 11 support.