UnsatisfiedLinkError on slaves after restart of master

[Stefan Drissen]

Jenkins: 1.627

TFS plug-in: 4.0.0

Java: jre1.8.0_60

Intermittently multiple slaves are failing to build with:

Started by user Stefan Drissen
[EnvInject] - Loading node environment variables.
Building remotely on <slave> in workspace c:\exact\jenkins\workspace\7.20\build work
FATAL: java.io.IOException: Remote call on <machine> failed
java.lang.RuntimeException: java.io.IOException: Remote call on <machine> failed
 at hudson.plugins.tfs.model.Server.execute(Server.java:110)
 at hudson.plugins.tfs.model.Project.extractChangesetNumber(Project.java:193)
 at hudson.plugins.tfs.model.Project.getRemoteChangesetVersion(Project.java:189)
 at hudson.plugins.tfs.model.Project.getRemoteChangesetVersion(Project.java:205)
 at hudson.plugins.tfs.TeamFoundationServerScm.recordWorkspaceChangesetVersion(TeamFoundationServerScm.java:262)
 at hudson.plugins.tfs.TeamFoundationServerScm.checkout(TeamFoundationServerScm.java:211)
 at hudson.model.AbstractProject.checkout(AbstractProject.java:1277)
 at hudson.model.AbstractBuild$AbstractBuildExecution.defaultCheckout(AbstractBuild.java:610)
 at jenkins.scm.SCMCheckoutStrategy.checkout(SCMCheckoutStrategy.java:86)
 at hudson.model.AbstractBuild$AbstractBuildExecution.run(AbstractBuild.java:532)
 at com.tikal.jenkins.plugins.multijob.MultiJobBuild$MultiJobRunnerImpl.run(MultiJobBuild.java:134)
 at hudson.model.Run.execute(Run.java:1741)
 at com.tikal.jenkins.plugins.multijob.MultiJobBuild.run(MultiJobBuild.java:73)
 at hudson.model.ResourceController.execute(ResourceController.java:98)
 at hudson.model.Executor.run(Executor.java:408)
Caused by: java.io.IOException: Remote call on <machine> failed
 at hudson.remoting.Channel.call(Channel.java:786)
 at hudson.plugins.tfs.model.Server.execute(Server.java:106)
 ... 14 more
Caused by: java.lang.UnsatisfiedLinkError: com.microsoft.tfs.jni.internal.platformmisc.NativePlatformMisc.nativeGetEnvironmentVariable(Ljava/lang/String;)Ljava/lang/String;

A restart of the slave (Windows 2008r2 - started as a Windows scheduled task) solves the issue (for a while). Why?

The 4.0.0 version of the TFS plug-in is supposed to be putting the tfs sdk in a place available for the slave - I have to admit that after searching the machine I cannot find it - but since a restart of the slave solves the issue, it must be somewhere.

The issue seems to manifest after a restart of the master Jenkins service (running on Windows 8.1 x64).

Best regards,

Stefan

[Peter McEvoy]

See exact same behaviour with Jenkins 2.7.1 and TFS plugin 5.0.0.0...   Updating plugins on master which needs to restart forces me to cycle all Jenkins agents...  Little worried that this is almost a year later and not getting more attention...

[Sinai Rijkov]

Same here too with Jenkins 1.6.52 . Windows Server 2008 R2 Standard.

only restart help to reconnect slave back to TFS. 

Thank you.