5. Trouble shooting

Q1. DAQ filesystems are 100% full

• The data disks on the DAQ are managed by CDR. Once a file is transferred, the CDR will run some cross checks to confirm that there were no transfer errors and that the file has been migrated to CASTOR. It will then purge the file from the local disk. Purging is handled by the purger script and in general, the file is not deleted immediately. If the local disks become full, this can mean either:

  The purger mechanism has been intentionally turned off

  The purger mechanism is on but the CDR will not purge files

If it is the former then the problem is easily fixed. If however the CDR blocks the deletion of local files, the most probable reason is that there has been an error in the transfer of the file to storage. The local file purge is then blocked until the discrepancy has been resolved. Users should under no circumstances start deleting files from the local disks of their own accord. Doing so may result in unrecoverable loss of data. cdr.support@cern.ch should be contacted .

Q2. I get error mail messages from Cron under the CDR account

In such a case cdr.support@cern.ch should be contacted .

Q3. There is a problem with the CDR service request form

In such a case cdr.support@cern.ch should be contacted .

Q4 How do I know if the CDR is running on my DAQ

When CDR is running on the DAQ, the stage_watchdog is executed every few minutes from crontab. The main CDR process is call stage_control. The UNIX ps command can be used to check for either of these processes.