1

When starting up a fresh instance of JBoss 6.1, I get the following error from Arjuna Recovery Manager, which I believe is part of the jboss-ts transaction service package. This error blocks the web server from starting up properly as the recoverymanager is a dependency for it.

Arjuna version 4.14.0 is used for this version of jboss.

Looking through the source, it seems there are no logging calls within that recoverymanager initialization aside from that fatal error, which makes debugging the issue even harder.

Here is the error stack trace of the issue:

09:14:02,946 ERROR [AbstractKernelController] Error installing to Create: name=RecoveryManager state=Configured: com.arjuna.ats.arjuna.exceptions.FatalError: ARJUNA-12359 SocketProcessId.getpid could not get unique port. at com.arjuna.ats.internal.arjuna.utils.SocketProcessId. SocketProcessId.java:82) [:6.1.0.Final]

... Java libraries [1.6.0_26]

at com.arjuna.ats.arjuna.utils.Utility.initDefaultProcess(Utility.java:336) [:6.1.0.Final] at com.arjuna.ats.arjuna.utils.Utility.getProcess(Utility.java:348) [:6.1.0.Final] at com.arjuna.ats.arjuna.utils.Utility.getpid(Utility.java:287) [:6.1.0.Final] at com.arjuna.ats.arjuna.common.Uid.(Uid.java:84) [:6.1.0.Final] at com.arjuna.ats.arjuna.StateManager.(StateManager.java:829) [:6.1.0.Final] at com.arjuna.ats.arjuna.StateManager.(StateManager.java:802) [:6.1.0.Final] at com.arjuna.ats.arjuna.coordinator.BasicAction.(BasicAction.java:65) [:6.1.0.Final] at com.arjuna.ats.arjuna.coordinator.TwoPhaseCoordinator.(TwoPhaseCoordinator.java:58) [:6.1.0.Final] at com.arjuna.ats.arjuna.AtomicAction.(AtomicAction.java:71) [:6.1.0.Final] at com.arjuna.ats.internal.arjuna.recovery.AtomicActionRecoveryModule.(AtomicActionRecoveryModule.java:263) [:6.1.0.Final] at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method) ... Java libraries [1.6.0_26]> at com.arjuna.ats.internal.arjuna.recovery.PeriodicRecovery.loadModule(PeriodicRecovery.java:893) [:6.1.0.Final]

EDIT: We found the culprit for this one. Our /etc/network/interfaces was broken, resulting in localhost loopback interface being down.

Broken:

auto lo, eth0

Fixed:

auto lo eth0

Lesson learned: Double check your server configurations, don't blame everything on JBoss. Most common solution provided for this issue is fixing /etc/hosts like Bart suggested. Unfortunately ours was due to another configuration issue.

Jukka Dahlbom
  • 111
  • 4

1 Answers1

1

Apparently, there was a bug that could lead to this in a previous version. Wouldn't hurt checking if the patch provided is incorporated in the current version.

You should also check out your hosts-file. This needs to be immaculate for Arjuna/JBoss to run. Make sure you have a line that looks like this:

10.12.34.56 jboss.mydomain.org jboss

Where you start with your own IP, followed by the FQDN and finally the hostname.

Bart De Vos
  • 17,911
  • 6
  • 63
  • 82
  • Thanks for the answer, looks like the /etc/hosts is related to this problem most often. However, this time it was not the case, and explicitly setting the IP in /etc/hosts was not necessary, and -as pointed out by our admin- is lot of extra work that is probably harmful for maintenance. What is correct about this answer is that the hosts file MUST be valid. If it IS broken, jboss probably won't run. – Jukka Dahlbom Oct 26 '11 at 07:10