HBASE-28647 Support streams in REST Client, RemoteHTable and RemoteAdmin #6010

stoty · 2024-06-20T14:57:51Z

No description provided.

Apache9

I do not fully understand what does the streams mean here...

All requests and responses are fully kept in memory here I think?

Apache9 · 2024-07-03T12:04:36Z

hbase-rest/src/main/java/org/apache/hadoop/hbase/rest/ProtobufHttpEntity.java

+ * For sending a Protobuf encoded object via Apache HttpClient efficiently without an interim byte
+ * array. This exposes the underlying Apache HttpClient types, but so do the other client classes.
+ */
+@InterfaceAudience.Public


Does this need to be IA.Public?

Yes, we expect the user to explicitly create these entities.

i have not updated the tests to use them, but the intentended usage is similar to how
org.apache.hadoop.hbase.rest.client.RemoteHTable.put(Put) is implemented.

At least have a section in the javadoc to show how to use this class?

And since we have not reach an agreement on whether to move RemoteAdmin and RemoteHTable to src/main, I'm not sure whether we should make this class IA.Public...

I only mentined RemoHTable as example where this is used, but this
is not limited to RemotHTable, it should also be used directly in user code.

I will add a standalone test case that uses this class as an illustration.

At least have a section in the javadoc to show how to use this class?

And since we have not reach an agreement on whether to move RemoteAdmin and RemoteHTable to src/main, I'm not sure whether we should make this class IA.Public...

This is not tied to RemoteAdmin / RemoteHTable.
This is just a more efficient way to use the existing public Client class (or any custom client based on Apache HttpClient)

The RemoteAdmin / RemoteHTable API does not even change, this new class only used internally in the RemoteHTable implementation.

Apache9 · 2024-07-03T12:05:07Z

hbase-rest/src/main/java/org/apache/hadoop/hbase/rest/ProtobufHttpEntity.java

+
+  @Override
+  public boolean isStreaming() {
+    // TODO Auto-generated method stub


Remove this line?

stoty · 2024-07-08T14:36:33Z

I do not fully understand what does the streams mean here...

All requests and responses are fully kept in memory here I think?

Currently they are.
The current code calls the Apache HttpClient getResponseBody() method, which will cause the client to wait untill all data is received, and load it into a byte array.

However, the goal is to avoid having to do that.

Protobuf primarily works on streams, so for a large resultset, we may reduce both processing (wall clock) time and memory consumption by not buffering the whole response into memory, but reading directly from the stream, so that

We do not have to wait for the full response to arrive before starting to process it.
We do not have to copy the whole response into a single byte array.
The processed response segments can be GCd while we are processing the rest of the message.

The Cell/Cellset structures are still kept in memory, but we avoid having to explicitly store them twice during processing (once the serialized byte array and once the java POJOs)

Apache9 · 2024-07-12T12:53:21Z

Please fix the javac and javadoc warnings if possible?

stoty · 2024-07-15T10:24:17Z

I have added org.apache.hadoop.hbase.rest.TestGetAndPutResource.testMultipleCellPutPBEntity() to illustrate how ProtobufEntity can be used to more efficiently execute PUT operations.

stoty · 2024-07-15T12:39:09Z

Fixed the warnings (though most were in code that I haven't touched)
and added a test that can serve as an example on using ProtobufEntity.
PTAL @Apache9 .

stoty · 2024-08-16T07:08:16Z

Can you take a look @Apache9 ?

stoty · 2024-09-10T05:49:35Z

Now that you're back from vacation and hopefully settled can you take a look at this again, @Apache9 ?

stoty · 2024-10-28T11:28:54Z

hbase-rest/src/test/java/org/apache/hadoop/hbase/rest/TestGetAndPutResource.java

+    path.append('/');
+    path.append("dummy_row");
+    HttpPut httpPut = new HttpPut(path.toString());
+    httpPut.setEntity(new ProtobufHttpEntity(model));


This is how a normal client can use ProtobufHttpEntity, @Apache9 .

Apache-HBase · 2025-02-19T15:18:13Z

🎊 +1 overall

Vote	Subsystem	Runtime	Logfile	Comment
+0 🆗	reexec	0m 30s		Docker mode activated.
-0 ⚠️	yetus	0m 3s		Unprocessed flag(s): --brief-report-file --spotbugs-strict-precheck --author-ignore-list --blanks-eol-ignore-file --blanks-tabs-ignore-file --quick-hadoopcheck
			_ Prechecks _
			_ master Compile Tests _
+1 💚	mvninstall	3m 16s		master passed
+1 💚	compile	0m 21s		master passed
+1 💚	javadoc	0m 17s		master passed
+1 💚	shadedjars	5m 58s		branch has no errors when building our shaded downstream artifacts.
			_ Patch Compile Tests _
+1 💚	mvninstall	3m 2s		the patch passed
+1 💚	compile	0m 21s		the patch passed
+1 💚	javac	0m 21s		the patch passed
+1 💚	javadoc	0m 16s		the patch passed
+1 💚	shadedjars	5m 52s		patch has no errors when building our shaded downstream artifacts.
			_ Other Tests _
+1 💚	unit	3m 50s		hbase-rest in the patch passed.
		24m 43s

Subsystem	Report/Notes
Docker	ClientAPI=1.43 ServerAPI=1.43 base: https://ci-hbase.apache.org/job/HBase-PreCommit-GitHub-PR/job/PR-6010/9/artifact/yetus-jdk17-hadoop3-check/output/Dockerfile
GITHUB PR	#6010
Optional Tests	javac javadoc unit compile shadedjars
uname	Linux 85fb0039742f 5.4.0-1103-aws #111~18.04.1-Ubuntu SMP Tue May 23 20:04:10 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux
Build tool	maven
Personality	dev-support/hbase-personality.sh
git revision	master / `a0803e6`
Default Java	Eclipse Adoptium-17.0.11+9
Test Results	https://ci-hbase.apache.org/job/HBase-PreCommit-GitHub-PR/job/PR-6010/9/testReport/
Max. process+thread count	1672 (vs. ulimit of 30000)
modules	C: hbase-rest U: hbase-rest
Console output	https://ci-hbase.apache.org/job/HBase-PreCommit-GitHub-PR/job/PR-6010/9/console
versions	git=2.34.1 maven=3.9.8
Powered by	Apache Yetus 0.15.0 https://yetus.apache.org

This message was automatically generated.

Apache-HBase · 2025-02-19T15:23:40Z

🎊 +1 overall

Vote	Subsystem	Runtime	Logfile	Comment
+0 🆗	reexec	0m 31s		Docker mode activated.
			_ Prechecks _
+1 💚	dupname	0m 0s		No case conflicting files found.
+0 🆗	codespell	0m 0s		codespell was not available.
+0 🆗	detsecrets	0m 0s		detect-secrets was not available.
+1 💚	@author	0m 0s		The patch does not contain any @author tags.
+1 💚	hbaseanti	0m 0s		Patch does not have any anti-patterns.
			_ master Compile Tests _
+1 💚	mvninstall	3m 17s		master passed
+1 💚	compile	0m 33s		master passed
+1 💚	checkstyle	0m 9s		master passed
+1 💚	spotbugs	0m 35s		master passed
+1 💚	spotless	0m 45s		branch has no errors when running spotless:check.
			_ Patch Compile Tests _
+1 💚	mvninstall	3m 3s		the patch passed
+1 💚	compile	0m 31s		the patch passed
+1 💚	javac	0m 31s		the patch passed
+1 💚	blanks	0m 0s		The patch has no blanks issues.
+1 💚	checkstyle	0m 9s		the patch passed
+1 💚	spotbugs	0m 41s		the patch passed
+1 💚	hadoopcheck	11m 42s		Patch does not cause any errors with Hadoop 3.3.6 3.4.0.
+1 💚	spotless	0m 43s		patch has no errors when running spotless:check.
			_ Other Tests _
+1 💚	asflicense	0m 10s		The patch does not generate ASF License warnings.
		30m 7s

Subsystem	Report/Notes
Docker	ClientAPI=1.43 ServerAPI=1.43 base: https://ci-hbase.apache.org/job/HBase-PreCommit-GitHub-PR/job/PR-6010/9/artifact/yetus-general-check/output/Dockerfile
GITHUB PR	#6010
Optional Tests	dupname asflicense javac spotbugs checkstyle codespell detsecrets compile hadoopcheck hbaseanti spotless
uname	Linux b9ac3c8ec6bf 5.4.0-1103-aws #111~18.04.1-Ubuntu SMP Tue May 23 20:04:10 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux
Build tool	maven
Personality	dev-support/hbase-personality.sh
git revision	master / `a0803e6`
Default Java	Eclipse Adoptium-17.0.11+9
Max. process+thread count	84 (vs. ulimit of 30000)
modules	C: hbase-rest U: hbase-rest
Console output	https://ci-hbase.apache.org/job/HBase-PreCommit-GitHub-PR/job/PR-6010/9/console
versions	git=2.34.1 maven=3.9.8 spotbugs=4.7.3
Powered by	Apache Yetus 0.15.0 https://yetus.apache.org

This message was automatically generated.

stoty · 2025-02-24T06:35:04Z

Can we get back to this patch, @Apache9 ?

To recap:

This is a perf optimization for the included REST Client, that avoids copying the full HTTP response into a byte array before feeding it into protobuf, in favor of using streams, which saves both memory and CPU.

ProtobufHttpEntity is a similar optimization for uploading large mutations, which again avoids storing the protobuf output to a byte array before uploading, and instead allows streaming the protobuf marshaller output directly to HTTP.

I have also updated the code to use ClosableHttpResponse instead of HttpResponse.

None of this affects the server code.

While this does expose some Apache HttpClient classes in the API, this is nothing new, as the current Client already does the same.