sonyxperiadev / gerrit-events

MIT License
47 stars 62 forks source link

Connections possibly left unclosed #57

Open vrtareg opened 8 years ago

vrtareg commented 8 years ago

Hi Team

We are using this library in one of the Jira PlugIns to connect to the Gerrit server to gather the reviwes information.

It looks like that in our environment library leaves some unclosed leftover connections which then affecting PlugIn functionality as server denies connection because of conn limit from Gerrit server side.

Is it possible that somewhere exception is not catched and connection left open?

2016-11-21 14:18:58,526 Caesium-1-4 DEBUG ServiceRunner [c.m.p.j.gerrit.multipleRepo.MultipleRepoConfigManagerImpl] ssh connection failed com.sonymobile.tools.gerrit.gerritevents.ssh.SshException: com.jcraft.jsch.JSchException: SSH_MSG_DISCONNECT: 7 Too many concurrent connections at com.sonymobile.tools.gerrit.gerritevents.ssh.SshConnectionImpl.connect(SshConnectionImpl.java:181) at com.sonymobile.tools.gerrit.gerritevents.ssh.SshConnectionFactory.getConnection(SshConnectionFactory.java:98) at com.sonymobile.tools.gerrit.gerritevents.ssh.SshConnectionFactory.getConnection(SshConnectionFactory.java:77) at com.sonymobile.tools.gerrit.gerritevents.ssh.SshConnectionFactory.getConnection(SshConnectionFactory.java:59) at com.meetme.plugins.jira.gerrit.multipleRepo.MultipleRepoConfigManagerImpl.createBareClone(MultipleRepoConfigManagerImpl.java:538) at com.meetme.plugins.jira.gerrit.multipleRepo.MultipleRepoConfigManagerImpl.synchronize(MultipleRepoConfigManagerImpl.java:622) at com.meetme.plugins.jira.gerrit.index.RevisionIndexer.updateIndexes(RevisionIndexer.java:223) at com.meetme.plugins.jira.gerrit.scheduling.SynchronizerLauncher.execute(SynchronizerLauncher.java:36) at com.atlassian.sal.jira.scheduling.JiraPluginScheduler$JobDescriptor.runJob(JiraPluginScheduler.java:110) at com.atlassian.sal.jira.scheduling.JiraPluginScheduler.runJob(JiraPluginScheduler.java:80) at com.atlassian.scheduler.core.JobLauncher.runJob(JobLauncher.java:153) at com.atlassian.scheduler.core.JobLauncher.launchAndBuildResponse(JobLauncher.java:118) at com.atlassian.scheduler.core.JobLauncher.launch(JobLauncher.java:97) at com.atlassian.scheduler.caesium.impl.CaesiumSchedulerService.launchJob(CaesiumSchedulerService.java:443) at com.atlassian.scheduler.caesium.impl.CaesiumSchedulerService.executeLocalJob(CaesiumSchedulerService.java:410) at com.atlassian.scheduler.caesium.impl.CaesiumSchedulerService.executeQueuedJob(CaesiumSchedulerService.java:388) at com.atlassian.scheduler.caesium.impl.CaesiumSchedulerService$1.consume(CaesiumSchedulerService.java:285) at com.atlassian.scheduler.caesium.impl.CaesiumSchedulerService$1.consume(CaesiumSchedulerService.java:282) at com.atlassian.scheduler.caesium.impl.SchedulerQueueWorker.executeJob(SchedulerQueueWorker.java:65) at com.atlassian.scheduler.caesium.impl.SchedulerQueueWorker.executeNextJob(SchedulerQueueWorker.java:59) at com.atlassian.scheduler.caesium.impl.SchedulerQueueWorker.run(SchedulerQueueWorker.java:34) at java.lang.Thread.run(Thread.java:745) Caused by: com.jcraft.jsch.JSchException: SSH_MSG_DISCONNECT: 7 Too many concurrent connections at com.jcraft.jsch.Session.read(Session.java:987) at com.jcraft.jsch.UserAuthPublicKey.start(UserAuthPublicKey.java:198) at com.jcraft.jsch.Session.connect(Session.java:463) at com.jcraft.jsch.Session.connect(Session.java:183) at com.sonymobile.tools.gerrit.gerritevents.ssh.SshConnectionImpl.connect(SshConnectionImpl.java:177) ... 21 more

Regards, Areg

rsandell commented 8 years ago

That could be possible, the connection is designed to try to keep the stream flowing as uninterrupted as possible. So if it looses connection somehow it tries to reconnect. If something lingers on on either end it would build up.

vrtareg commented 8 years ago

Any way to prevent this and close unused connections?

rsandell commented 8 years ago

You would need to find it first :) then it would hopefully be pretty obvious how to fix the leak if you find where it is.

vrtareg commented 8 years ago

Any recommendations what to monitor and where to put debug output? I will be glad to do if I know how.

rsandell commented 8 years ago

The connect logic is the last line of the stack trace you provided. The main loop is here https://github.com/sonyxperiadev/gerrit-events/blob/master/src/main/java/com/sonymobile/tools/gerrit/gerritevents/GerritConnection.java#L314

vrtareg commented 7 years ago

So something is going wrong and new connection is created without closing current one. This is really killing as Gerrit servers has a limit on incoming connections from clients and on a next round it just denies new connection. Only way to fix this is to restart Jira instance ...

rsandell commented 7 years ago

Which is strange because if a connection fails to be established there shouldn't be any connection available to close.