Java VM terminated with a non-zero exit code (rc = -337)

[Former Member]

Hi,

It seems that our portal (7.0) was unavailable for five minutes (all users experienced problems which after a few minutes solved itself). We had a look at the logs, and dev_server0 shows that the server node did shut down and start again:

[Thr 6688] Wed Aug 13 10:19:36 2014

[Thr 6688] JLaunchIExitJava: exit hook is called (rc = -337)

[Thr 6688] **********************************************************************

[Thr 6688] *** ERROR => The Java VM terminated with a non-zero exit code.

[Thr 6688] *** Please see SAP Note 943602 , section 'J2EE Engine exit codes'

[Thr 6688] *** for additional information and trouble shooting.

[Thr 6688] **********************************************************************

[Thr 6688] JLaunchCloseProgram: good bye (exitcode = -337)

The referenced note doesn't offer much help. I've tried to find more info about rc -337, and the only info I've found is that rc -337 is: "Exiting listener loop - restart the node (MSLib)."

Do you guys know what exiting listener loop means, or what the rc -377 is?

Thanks!

M.

[hemanth2 · Accepted solution]

Hi Moye,

Hope you are doing good.

Such issues can also be caused due to a network failure. Can you check the below notes along with the network administrator to ascertain that there are no network related issues presently:

#124562     Hostname resolution problems (DNS timeouts)

#500235     Network Diagnosis with NIPING

#413330     Network timeout

If you run the niping tool at the time the issue occurred, it should check the whole network and report the error.

It is also possible to optimize the j2ee server in the manner it interacts with the message server for better performance. As Divyanshu mentioned, the note 1658385 - J2EE System fails with "exitcode = -334" explains, you need to optimize the manner in which the message server reconnects with the Java instances.

This can happen sometimes if you have quite a few nodes in cluster. To make sure the issue does not happen again, you need to increase for each server node the timeout it waits for the confirmation,

that is the value of 'ms.confirmation.timeout' property of ClusterManager along with the other parameters mentioned in the note. Also add the below parameters to the SCS instance file:

<sid>_SCS01_<host> in folder /usr/sap/<sid>/SYS/profile

ms/max_clients = 600

ms/max_open_requests = 2000

ms/audit=1

and

jstartup/keep_old_logfiles=3

The latter 2 parameters will prevent  overwritting log files in case of restart. Once the values are increased, the time for the timeout should increase enabling the server to connect to the message server.

Thank you!

_____________

Kind Regards,

Hemanth

SAP AGS
 

[Former Member]

Hi,

Thanks for helpful answers. I believe the problem was due to network issues.

M.

[hemanth2]

Great news Moye

[Sriram2009]

Hi Moye

If possible to do the full restart and then check the Java system. as mention in the SAP Note

1916706 - ERROR - The Java VM terminated with a non-zero exit code

BR

SS

[hemanth2]

Hi Sriram, this note (1916706) is not going to help a bit. The issue is with SCS and j2ee server node connection: rc -337.

Return code 337 usually means that the server/dispatcher process is restarted because it exits the   listener loop.

Kind Regards,

Hemanth

[divyanshu_srivastava3]

Hi Moye,

Can you share dev_server0/dispatcher and their std* logs ?

We have faced this issue once on our portal and this was because of some network latency, as far as I can recall. You can check the file before the restart of the node for high GC usage and any other possible error. Also, have a look at default trace for some hints.

Refer: 1622083 - J2EE System fails with "exitcode = -337"

Regards,

Divyanshu