One Star

MetaServletCaller times out its query, external scheduler shows fail

I've tried to get help from Talend support for this.
We use Jenkins as a scheduler, calling MetaServletCaller on Talend server, pointing to an url at localhost:8080.

The issue is that some internal Talend component fails (times out?) with its status query, returning an error to MetaServletCaller, which in turn is picked up by the scheduler, e.g.:
{"errorStatus":"CONNECTION_TO_SERVER_FAILED","executionTime":{"millis":2411386,"seconds":2411},"returnCode":0,"status":"UNKNOWN"}
However the job executes successfully. There is a related Tomcat failure also, with
2014-10-29 03:41:33 ERROR ExecutionTaskRefresher - Execution task state recovering failed
org.talend.administrator.common.exception.SystemException: org.talend.remote.commons.exception.ClientServerException: Time out server response for command ', startDate=null, endDate=null, idJobExecution=20141029030134_Rm0UJ, idJob=20141028_030127_J2dJB, remoteActionStatus=null, messageType=GET_JOB_EXECUTION_STATE, idRequestResponsePair=aF4FyfW2, message=null'
Can anyone tell me what is being queried here? I suspect the MySQL database (also on Talend server), but I cannot be sure.
These failures occur relatively often and cause downstream jobs not to start.
2 REPLIES
One Star

Re: MetaServletCaller times out its query, external scheduler shows fail

Can you give more info about the component and how you use them?
One Star

Re: MetaServletCaller times out its query, external scheduler shows fail

Jenkins executes a job remotely on Talend server, e.g.
$cmd = TAC_CALL.' -json-params=\'{"actionName":"runTask","authPass":"'.TAC_PASS.'","authUser":"'.TAC_USER.'","taskId":"'.$taskId.'","mode":"'.$mode.'","context":'.json_encode($context).'}\''
and monitors it via RESTful (MetaServlet).
Configuration for TAC_CALL:
define("TAC_URL",  "");
define("TAC_CALL", TAC_PATH.'WEB-INF/classes/MetaServletCaller.sh -url '.TAC_URL);
Talend support thinks it's a problem with connectivity between TAC and Jobserver, but I don't think so. Other jobs run nicely, it is just that occassionally the monitoring times out and returns the error to the scheduler. Looking at it from TAC, though, shows it finished without any issues.