HypericHQ_4U
Contributor
Contributor

Hyperic HQ 5.7.1 - Agent Fails to Start

Hyperic HQ 5.7.1

OS: IBM AIX 6.1, Level 6100-09

--------------------------------

We have seen this problem with Hyperic HQ Agent 4.6.6 and now 5.7.1 after upgrading. When starting the Hyperic HQ Agent using /bin/, the following error occurs:

--== [user][!]==-->./hq-agent.sh start


-n Starting HQ Agent

-n.

-n.

-n.

Remove stale pid file:/opt/hyperic-hq-agent-5.7.1/wrapper/sbin/../../wrapper/hq-agent.pid

WARNING: HQ Agent may have failed to start

------------------------------

No entries are written into either the wrapper.log or agent.log. I have noticed that when increasing the timeout for the wrapper startup and then restarting the agent, I can tell that a hq-agent.pid is created with a PID number inside the file, but the PID itself does is not shown as a currently running process (this should be obvious since the agent does not start at all).

When enabling debug logging through the bundles/agent-5.7.1/conf/wrapper.conf, there is still no information sent to the logs upon start up. When attempting to run the service wrapper individually (to confirm the wrapper file can be read), I get the following:

--== You are in /opt/hyperic-hq-agent-5.7.1/wrapper/sbin ==--

--== [user][!]==--> ./wrapper-aix-ppc-32

Segmentation fault (core dumped)

I believe that the issue is surrounding the wrapper, but I am not that experienced within the Hyperic programming to determine what should be done next. If there is any input out there, respond to this post.  Thanks

Eric

0 Kudos
5 Replies
admin
Immortal
Immortal

Hi,

Try to install NO JRE hyperic agent 5.7.1 from scratch and start it. Make sure you're using IBM Java.

Good Luck

0 Kudos
HypericHQ_4U
Contributor
Contributor

Hi Boris_Erblat,

Sorry for the very late response. I installed the NO JRE version of Hyperic Agent 5.7.1 and still received the same errors as in the original post. I used the following Java versions:

-Java 1.6.0 SR 14 (32 and 64-bit)

-Java 1.7.0 SR5 (32 and 64-bit)

I believe that the problem is more isolated to the wrapper, especially since the bundled JRE has been isolated from the equation. I'm just stuck on how to resolve it.

0 Kudos
admin
Immortal
Immortal

Hi,

A few questions:

  • What did you change (e.g updates) in your machine before that issue happened ?
  • Before start a new agent from scratch, did you remove previous agent, did you check  that there is no any process running of hyperic agent, did you delete hyperic's files in /tmp directory?
  • Do you have another AIX to verify your problem ?

Please check it and try again.

Another option, once you're using open source install hyperic-hq-agent -ppc-aix-5.7.1.

Good luck

0 Kudos
HypericHQ_4U
Contributor
Contributor

Boris,

My apologies for the late reply, I had to obtain information from our infrastructure admins who maintain the overall servers:

  • What did you change (e.g updates) in your machine before that issue happened ?

         The infrastructure team upgraded the OS level of AIX from 6100-08-03 to 6100-09-01. Within this upgrade, there were about 300 fixes that were applied to the system. Before the OS

          level upgrade, the agents would operate fine.

  • Before start a new agent from scratch, did you remove previous agent, did you check  that there is no any process running of hyperic agent, did you delete hyperic's files in /tmp directory?

        The agents were removed from the system, lingering processes were removed, and the /tmp directory was free of all Hyperic files.

  • Do you have another AIX to verify your problem ?

         As stated in the first question, before the OS level upgrade, the agents would operate fine under 6100-08-03.

Question: other than configuring logging within agent.conf, where else could I enable detailed logging? When the agent starts (and fails), no messages are written to the agent.log. I know that I could probably enable detailed logging for the wrapper in order to determine what the issue is.

0 Kudos
HypericHQ_4U
Contributor
Contributor

Just to update my notes above, I confirmed that starting up the Hyperic HQ Agent works when using the hq-agent-no-wrapper.sh file. I can start and stop the agent successfully.

This definitely isolates this issue being with the wrapper portion of the agent in correlation with the AIX version we have.

0 Kudos