ZEPPELIN-3411 Long running logic inside synchronized block in InterpreterSettingManager by jongyoul · Pull Request #2935 · apache/zeppelin

jongyoul · 2018-04-22T13:21:57Z

What is this PR for?

Removing redundant synchronized code to avoid blocking other logics.

What type of PR is it?

[Bug Fix]

Todos

- Change synchronized block to read/write lock

What is the Jira issue?

https://issues.apache.org/jira/browse/ZEPPELIN-3411

How should this be tested?

Current tests should be passed

Screenshots (if appropriate)

Questions:

Does the licenses files need update? No
Is there breaking changes for older versions? No
Does this needs documentation? No

jongyoul · 2018-04-22T13:22:40Z

Test passed. https://travis-ci.org/jongyoul/zeppelin/builds/367566398

jongyoul · 2018-04-22T13:23:22Z

-        //clean up metaInfos
-        intpSetting.setInfos(null);
-        copyDependenciesFromLocalPath(intpSetting);
-        intpSetting.closeInterpreters(user, noteId);


@zjffdu It was the original problem.

jongyoul · 2018-04-22T13:24:56Z

This is not a critial issue when running normally. But in case of problemistic situation like restarting forcibly to stop long running job of SparkInterpreter, some logic hangs inside synchronized block.

zjffdu · 2018-04-22T13:44:11Z

Do you have jstack so that I can understand what the exact problem is ?

jongyoul · 2018-04-22T13:53:43Z

@zjffdu Let me make it

jongyoul · 2018-04-23T03:18:51Z

@zjffdu I've received this issue report from one of my colleagues that he ran a wrong code to read all of the data from a large directory and it took for 20mins with a spark job and he tried to stop a spark interpreter. and Then he tried to create a new note, but he couldn't see an interpreter lists from creating note page and a list-interpreter when clicking a button on a note.

BUT I've tried to make it with current master into my local but it fails because 'restart' always work fine. It looks only until 0.7.3.

Generally, this code doesn't a problem anymore in current master and branch-0.8, but this kind of synchronized blocks don't look good. I prefer to change it but it's not a bug now. What do you think of it?

zjffdu · 2018-04-23T03:59:44Z

   */
  private final Map<String, InterpreterSetting> interpreterSettings =
      Maps.newConcurrentMap();
+  private final ReentrantReadWriteLock interpreterSettingsLock = new ReentrantReadWriteLock();


Actually interpreterSettings is ConcurrentHashMap, I think we don't need extra lock here.

Yes, I thought it and wanted to confirm it. Then, I'll remove all of the locks related to interpreterSettings in this class.

…s implemented by currentHashMap

jongyoul · 2018-04-23T06:35:04Z

Removed all of the synchronized blocks from InterpreterSettingManager for interpreterSettings

zjffdu · 2018-04-23T06:39:46Z

-        closeThreads.add(t);
-      }
+    Collection<InterpreterSetting> intpSettings = interpreterSettings.values();
+    for (final InterpreterSetting intpSetting : intpSettings) {


Merge them into one line

for (InterpreterSetting intpSetting : interpreterSettings.values()) {

zjffdu · 2018-04-23T06:41:27Z

 import com.google.gson.Gson;
 import com.google.gson.GsonBuilder;
 import com.google.gson.reflect.TypeToken;
+import java.util.concurrent.locks.ReentrantReadWriteLock;


this could be removed

zjffdu · 2018-04-23T06:42:07Z

THanks @jongyoul Just several minor issues, rest look good to me.

Simplified `for` loop

jongyoul · 2018-04-23T06:52:22Z

After finishing CI, I'll merge it

jongyoul · 2018-04-23T08:12:00Z

Test passed. https://travis-ci.org/jongyoul/zeppelin/builds/369961193 Will merge it

…eterSettingManager ### What is this PR for? Removing redundant synchronized code to avoid blocking other logics. ### What type of PR is it? [Bug Fix] ### Todos * [x] - Change synchronized block to read/write lock ### What is the Jira issue? * https://issues.apache.org/jira/browse/ZEPPELIN-3411 ### How should this be tested? * Current tests should be passed ### Screenshots (if appropriate) ### Questions: * Does the licenses files need update? No * Is there breaking changes for older versions? No * Does this needs documentation? No Author: Jongyoul Lee <jongyoul@gmail.com> Closes apache#2935 from jongyoul/ZEPPELIN-3411 and squashes the following commits: 3b90155 [Jongyoul Lee] Removed unused `import` statements Simplified `for` loop 4691301 [Jongyoul Lee] Removed lock/synchronized codes because interpreterSettings already is implemented by currentHashMap 24be692 [Jongyoul Lee] Removed all synchronized blocks and replace them to read/write lock

weand · 2018-04-23T22:15:53Z

are you going to merge that into branch-0.8 as well?

jongyoul · 2018-04-23T23:39:00Z

@weand Oops. I was confused with another issue. I’ll cherry-pick this into 0.8

…eterSettingManager ### What is this PR for? Removing redundant synchronized code to avoid blocking other logics. ### What type of PR is it? [Bug Fix] ### Todos * [x] - Change synchronized block to read/write lock ### What is the Jira issue? * https://issues.apache.org/jira/browse/ZEPPELIN-3411 ### How should this be tested? * Current tests should be passed ### Screenshots (if appropriate) ### Questions: * Does the licenses files need update? No * Is there breaking changes for older versions? No * Does this needs documentation? No Author: Jongyoul Lee <jongyoul@gmail.com> Closes #2935 from jongyoul/ZEPPELIN-3411 and squashes the following commits: 3b90155 [Jongyoul Lee] Removed unused `import` statements Simplified `for` loop 4691301 [Jongyoul Lee] Removed lock/synchronized codes because interpreterSettings already is implemented by currentHashMap 24be692 [Jongyoul Lee] Removed all synchronized blocks and replace them to read/write lock

…eterSettingManager Removing redundant synchronized code to avoid blocking other logics. [Bug Fix] * [x] - Change synchronized block to read/write lock * https://issues.apache.org/jira/browse/ZEPPELIN-3411 * Current tests should be passed * Does the licenses files need update? No * Is there breaking changes for older versions? No * Does this needs documentation? No Author: Jongyoul Lee <jongyoul@gmail.com> Closes apache#2935 from jongyoul/ZEPPELIN-3411 and squashes the following commits: 3b90155 [Jongyoul Lee] Removed unused `import` statements Simplified `for` loop 4691301 [Jongyoul Lee] Removed lock/synchronized codes because interpreterSettings already is implemented by currentHashMap 24be692 [Jongyoul Lee] Removed all synchronized blocks and replace them to read/write lock (cherry picked from commit 6559c5e) Change-Id: Ic28a3e9cf8ec93c84596861716c3cc25dedd96f5

…eterSettingManager ### What is this PR for? Removing redundant synchronized code to avoid blocking other logics. ### What type of PR is it? [Bug Fix] ### Todos * [x] - Change synchronized block to read/write lock ### What is the Jira issue? * https://issues.apache.org/jira/browse/ZEPPELIN-3411 ### How should this be tested? * Current tests should be passed ### Screenshots (if appropriate) ### Questions: * Does the licenses files need update? No * Is there breaking changes for older versions? No * Does this needs documentation? No Author: Jongyoul Lee <jongyoul@gmail.com> Closes apache#2935 from jongyoul/ZEPPELIN-3411 and squashes the following commits: 3b90155 [Jongyoul Lee] Removed unused `import` statements Simplified `for` loop 4691301 [Jongyoul Lee] Removed lock/synchronized codes because interpreterSettings already is implemented by currentHashMap 24be692 [Jongyoul Lee] Removed all synchronized blocks and replace them to read/write lock

Removed all synchronized blocks and replace them to read/write lock

24be692

jongyoul commented Apr 22, 2018

View reviewed changes

zjffdu reviewed Apr 23, 2018

View reviewed changes

Removed lock/synchronized codes because interpreterSettings already i…

4691301

…s implemented by currentHashMap

zjffdu reviewed Apr 23, 2018

View reviewed changes

Removed unused import statements

3b90155

Simplified `for` loop

asfgit closed this in 75dd195 Apr 23, 2018

Conversation

jongyoul commented Apr 22, 2018

What is this PR for?

What type of PR is it?

Todos

What is the Jira issue?

How should this be tested?

Screenshots (if appropriate)

Questions:

Uh oh!

jongyoul commented Apr 22, 2018

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jongyoul commented Apr 22, 2018

Uh oh!

zjffdu commented Apr 22, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jongyoul commented Apr 22, 2018

Uh oh!

jongyoul commented Apr 23, 2018

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jongyoul commented Apr 23, 2018

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zjffdu commented Apr 23, 2018

Uh oh!

jongyoul commented Apr 23, 2018

Uh oh!

jongyoul commented Apr 23, 2018

Uh oh!

weand commented Apr 23, 2018

Uh oh!

jongyoul commented Apr 23, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

zjffdu commented Apr 22, 2018 •

edited

Loading