To debug baseline update failures, examine the Dgraph request log and the baseline update log first, followed by the EAC process logs.
Use the following recommendations:
Review Dgraph request logs. Review the logs around the time of the baseline update failure, to rule out issues in the Dgraph.
Notice the times when health checks were sent to the Dgraph, the Dgraph was restarted, the partial updates were issued, and the last query was issued.
For example, this modified abstract from the Dgraph request log shows activity for a period of time:
12096521815/1/09 14:29 last search query 12096522265/1/09 14:30 health check 12096526095/1/09 14:36 last health check for x time 12096571605/1/09 15:52 health checks resume 12096574435/1/09 15:57 last empty health check 12096601195/1/09 16:41 Dgraph startup 12096601435/1/09 16:42 first query
Notice that the Dgraph did not receive any requests besides health checks for a period of time from 14:29 to 15:57. The log does not include error messages. The Dgraph was not restarted during this time. These observations indicate that the problem that led to the baseline update failure in this example possibly occurred outside of the Dgraph.
Review baseline update.out logs. For example, in the case below, observe that an error occurred while stopping the Dgraph component:
[05.01.09 10:07:54] INFO: Stopping component 'Dgraph1'. [05.01.09 10:17:54] SEVERE: Error communicating with EAC agent while stopping component. Occurred while executing line 5 of valid BeanShell script:
To investigate further the reason for why the EAC was not able to stop the Dgraph component, examine the logs for EAC processes and increase their verbosity.
Increase the verbosity of the EAC process logs.
Specify the EAC logging configuration in
[ENDECA_CONF]
/conf/logging.propertiescom.endeca.eac.invoke
,com.endeca.eac.process
andcom.endeca.eac.main
toFINE
. This provides additional debug information, if the baseline update process fails again.Note
Monitor the size of the
ENDECA_CONF
/logsTo continue with the example, the EAC process logs may, for instance, indicate an outage of a hardware component between the Web server and the Dgraph server. These logs may further assist you if the baseline update fails again.