Auto MC Potential or Actual Problems

Unable to get a grid proxy or store it in the myproxy server.

An error message such as this appears:
Warning: your certificate and proxy will expire Wed Dec 20 05:57:12 2006
which is within the requested lifetime of the proxy
grid-proxy-init failed
The solution is to remove any files owned by the user in /tmp, log out, log in again, and retry the setup script.

Unable to submit job

Appears in logs and in email message. Caused by:

• Expired proxy
  • Check with "$ myproxy-info -d" .
  • Obtain a new one which can be done by sourcing the setup file.
• Not in grid-mapfiles or DZero VOMS
  • Error message in logs is:
    ERROR: Failed to connect to local queue manager
    AUTHENTICATE:1003:Failed to authenticate with any method
    AUTHENTICATE:1004:Failed to authenticate using GSI
    GSI:5004:Failed to get authorization from server. Either the server does not trust your certificate, or you are not in the server's authorization file (grid-mapfile)
  • A Kerberos generated proxy should not see this error. If using a grid certificate generated proxy send the grid subject to parag at fnal.gov asking for inclusion in the Samgrid grid-mapfiles. A submitter must also register their grid subject in the DZero VOMS please see http://www-d0.fnal.gov/VO/DZero_VO_Instructions.html for instructions.
• Unable to get a new request
  • Maybe a SAM problem see below.
  • Check MC production page links to current requests to see if requests are available. Note the pages are updated once an hour so the they may show requests available when there really aren't any. If there are none available report it to the MC production list. Continue running the daemon until new requests are available.
• SAM problems
  • Notify an expert via SAM issue tracker.
• SAMGrid problems
  • Notify an expert via SAM issue tracker.

Excessive number of grid jobs for a request

Job number on status page is greater than 9. Caused by:

• Bad Request
  • Notify an expert via MC production list.
• Problems at execution site
  • Notify an expert via MC production list.

Job stays idle never goes to running status

Job status on status page is idle for more than a hour. Caused by:

• Execution site down
  • Notify an expert via MC production list.
• Incorrect Job Description File
  • Notify an expert via MC production list.

Useful Links

• Dial-A-Job Status
• SAMGrid list of resources
• SAMGrid jobs at the scheduler
• Monte carlo production page
• SAM At A Glance
• Auto MC Documentation