ZOOKEEPER-3706: ZooKeeper.close() would leak SendThread when the netw… by yfxhust · Pull Request #1235 · apache/zookeeper

yfxhust · 2020-01-25T08:38:23Z

…ork is broken

add unit test to verify the bug
bypass the SendThread.startConnect() by throw RuntimeExcepth if state.isAlive is false

Author: Fangxi Yin yinfangxi@kuaishou.com

maoling · 2020-01-31T08:23:18Z

@yfxhust
Great issue description in JIRA. this issue is a little hidden. Do you observe it in the production?

yfxhust · 2020-02-03T03:47:50Z

@yfxhust
Great issue description in JIRA. this issue is a little hidden. Do you observe it in the production?

We observed occasional SendThread leak case in our production. But the root cause is not 100 percent sure. But from source code and some network clues, we believe that this issue is a potential cause.

hanm · 2020-02-03T22:40:16Z

nice work - left two comments. since the request timeout is a feature disabled by default, the impact of this bug is limited to those who enabled this feature only.

yfxhust · 2020-02-04T04:12:40Z

nice work - left two comments. since the request timeout is a feature disabled by default, the impact of this bug is limited to those who enabled this feature only.

Thanks for your comments

hanm · 2020-02-05T05:14:10Z

left two more comments - current approach should work but I am wondering if my proposal would sound better in term of more structurally solve the race condition around the state variable; also was a little bit concerning around RuntimeException we throw and the implications it brought (w.r.t resource leaks)

yfxhust · 2020-02-06T09:52:39Z

left two more comments - current approach should work but I am wondering if my proposal would sound better in term of more structurally solve the race condition around the state variable; also was a little bit concerning around RuntimeException we throw and the implications it brought (w.r.t resource leaks)

Thank you for your comments. I adapt my code with your suggestions. Please review the latest patch.

maoling

@hanm Do you have other concerns?

hanm · 2020-02-19T05:30:26Z

I just have two nits - one on the sync approach and the other on unit test.

yfxhust · 2020-02-19T12:09:15Z

I just have two nits - one on the sync approach and the other on unit test.

Thank you for your comments. Adapt my code with your suggestion. Please see the latest patch.

yfxhust · 2020-02-25T09:10:04Z

@hanm Any further comments ?

hanm

LGTM.

Any feedback from other committers? If not, I'll land this in next few days.

eolivelli

Awesome work.
I left a few comments

yfxhust · 2020-03-02T04:55:08Z

@eolivelli Thank you for your comments. I adapt the patch with your suggestions. Please review the latest patch.

…ork is broken - add unit test to verify the bug - wrap state modification in SendThread.changeZkState() - SendThread.startConnect() could be interrupted by IOException caused by SendThread.changeZkState() Author: Fangxi Yin <yinfangxi@kuaishou.com>

eolivelli

LGTM

@hanm please go ahead and merge if you are okay with the latest version

hanm · 2020-03-03T02:15:11Z

merged to master - thanks for driving this @yfxhust

…ork is broken - add unit test to verify the bug - bypass the SendThread.startConnect() by throw RuntimeExcepth if state.isAlive is false Author: Fangxi Yin <yinfangxikuaishou.com> Author: yinfangxi <yinfangxi@kuaishou.com> Reviewers: Michael Han <hanm@apache.org>, Enrico Olivelli <eolivelli@gmail.com>, maoling Closes apache#1235 from yfxhust/ZOOKEEPER-3706

…ork is broken - add unit test to verify the bug - bypass the SendThread.startConnect() by throw RuntimeExcepth if state.isAlive is false Author: Fangxi Yin <yinfangxikuaishou.com> Author: yinfangxi <yinfangxi@kuaishou.com> Reviewers: Michael Han <hanm@apache.org>, Enrico Olivelli <eolivelli@gmail.com>, maoling Closes #1235 from yfxhust/ZOOKEEPER-3706

…ork is broken - add unit test to verify the bug - bypass the SendThread.startConnect() by throw RuntimeExcepth if state.isAlive is false Author: Fangxi Yin <yinfangxikuaishou.com> Author: yinfangxi <yinfangxi@kuaishou.com> Reviewers: Michael Han <hanm@apache.org>, Enrico Olivelli <eolivelli@gmail.com>, maoling Closes apache#1235 from yfxhust/ZOOKEEPER-3706

maoling reviewed Jan 25, 2020

View reviewed changes

Comment thread zookeeper-server/src/test/java/org/apache/zookeeper/ClientCnxnSocketFragilityTest.java

maoling reviewed Jan 25, 2020

View reviewed changes

Comment thread zookeeper-server/src/test/java/org/apache/zookeeper/ClientCnxnSocketFragilityTest.java

maoling reviewed Jan 31, 2020

View reviewed changes

Comment thread zookeeper-server/src/main/java/org/apache/zookeeper/ClientCnxn.java

Comment thread zookeeper-server/src/main/java/org/apache/zookeeper/ClientCnxn.java Outdated

yfxhust force-pushed the ZOOKEEPER-3706 branch from 30c07d2 to ce347e0 Compare February 3, 2020 06:10

hanm requested changes Feb 3, 2020

View reviewed changes

Comment thread zookeeper-server/src/main/java/org/apache/zookeeper/ClientCnxn.java Outdated

Comment thread zookeeper-server/src/main/java/org/apache/zookeeper/ClientCnxn.java Outdated

yfxhust force-pushed the ZOOKEEPER-3706 branch from ce347e0 to bf86062 Compare February 4, 2020 04:48

yfxhust force-pushed the ZOOKEEPER-3706 branch from bf86062 to 74975f3 Compare February 6, 2020 09:42

yfxhust force-pushed the ZOOKEEPER-3706 branch 3 times, most recently from f787f2b to 95accad Compare February 10, 2020 08:18

maoling reviewed Feb 15, 2020

View reviewed changes

Comment thread zookeeper-server/src/main/java/org/apache/zookeeper/ClientCnxn.java Outdated

Comment thread zookeeper-server/src/main/java/org/apache/zookeeper/ClientCnxn.java

Comment thread zookeeper-server/src/main/java/org/apache/zookeeper/ClientCnxn.java

yfxhust force-pushed the ZOOKEEPER-3706 branch 2 times, most recently from d1e235b to 16500a6 Compare February 17, 2020 03:48

maoling approved these changes Feb 17, 2020

View reviewed changes

hanm reviewed Feb 19, 2020

View reviewed changes

Comment thread zookeeper-server/src/test/java/org/apache/zookeeper/ClientCnxnSocketFragilityTest.java Outdated

yfxhust force-pushed the ZOOKEEPER-3706 branch from 16500a6 to 52b2ef7 Compare February 19, 2020 11:24

yfxhust force-pushed the ZOOKEEPER-3706 branch from 52b2ef7 to 8938807 Compare February 19, 2020 12:12

hanm approved these changes Feb 28, 2020

View reviewed changes

eolivelli requested changes Feb 28, 2020

View reviewed changes

yfxhust force-pushed the ZOOKEEPER-3706 branch from 8938807 to a630b12 Compare March 2, 2020 04:52

yfxhust force-pushed the ZOOKEEPER-3706 branch from a630b12 to c06eaa0 Compare March 2, 2020 04:58

eolivelli approved these changes Mar 2, 2020

View reviewed changes

asfgit closed this in d3e72c6 Mar 3, 2020

lhotari mentioned this pull request Jan 27, 2022

Bump to Pulsar 2.8.2 apache/pulsar-helm-chart#190

Closed

Conversation

yfxhust commented Jan 25, 2020

Uh oh!

Uh oh!

Uh oh!

maoling commented Jan 31, 2020

Uh oh!

Uh oh!

Uh oh!

yfxhust commented Feb 3, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

hanm commented Feb 3, 2020

Uh oh!

yfxhust commented Feb 4, 2020

Uh oh!

hanm commented Feb 5, 2020

Uh oh!

yfxhust commented Feb 6, 2020

Uh oh!

Uh oh!

Uh oh!

Uh oh!

maoling left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

hanm commented Feb 19, 2020

Uh oh!

yfxhust commented Feb 19, 2020

Uh oh!

yfxhust commented Feb 25, 2020

Uh oh!

hanm left a comment

Choose a reason for hiding this comment

Uh oh!

eolivelli left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

yfxhust commented Mar 2, 2020

Uh oh!

eolivelli left a comment

Choose a reason for hiding this comment

Uh oh!

hanm commented Mar 3, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

yfxhust commented Feb 3, 2020 •

edited

Loading