Don't Save Run Record When Triggering Graph from Simple HTTP API

mukunku

• November 07, 2017 00:00

• Answered

Follow

Hello,

I'm triggering my graphs using the graph_run operation and was wondering if there's any parameter that can be passed so as to prevent a run record from being generated.

Some graphs are called very frequently and I don't want them generating tons of logs. Preferably I would only want a run record to be saved if the graph fails.

Thanks,
Sal

• 

  Vladimir Barton

  • November 13, 2017 12:03

  0

  Comment actions Permalink

  Hi Sal,
  there are a couple of options that you can take advantage of when trying to achieve a lower amount of log data:
  
  

  • As you rightly indicated, it can be achieved by adding a log level parameter directly to the URL when calling a ‘graph_run’ operation. By default, the logging level is set to INFO on all graphs. If you needed to log errors only, you could set the ‘config_logLevel’ parameter to ERROR. Example URL: http://localhost:8080/clover/request_pr ... evel=ERROR
    
  
  
  • Another option is to set the log level property on the graph itself. In the server console, you would navigate to Sandboxes and select a desired graph within a sandbox. On the Config properties tab > Create New Config Property section, select the ‘log_level’ property in the dropdown, populate the respective field on the right with “ERROR” and click Add.
    Note: With both of these approaches, job logs of successfully finished graphs will be empty (run logs will only contain header information such as project, graph, run ID etc.) On the contrary, if the graph failed, the run log would consist of relevant error details.
    
  
  
  • If you aim at decreasing the amount of log data in general, another option would be to schedule archivator tasks so that they run more often and maintain shorter history of logs. This can be done within the Scheduling tab > Triggered task: Archivator.

  
  
  Kind regards,

Please sign in to leave a comment.