PETAL: 3.4.2 Scheduler Monitoring

The QStar software is provided with a scheduler that acts in the same way crontab works under Linux. Even the notation used to indicate the timings of execution of the job is taken directly from the crontab notation. Apart from the native commands of the QStar software, any command or script can be assigned to a job and scheduled in the same manner.

In the actual configuration of the PETAL Archive, the scheduler is set up to run the following jobs:

  • Incremental copy of current tape for the volume deep2 scheduled daily at 7.00 AM and 6.00 PM
  • Incremental copy of current tape for the volume near2 scheduled daily at 6.00 AM and 8.00 PM
  • Flushing of the cache for the volume Deep1 scheduled daily at 11.55 PM
  • Flushing of the cache for the volume Deep2 scheduled daily at 00.55 AM
  • Flushing of the cache for the volume Deep1 scheduled daily at 01.55 AM
  • Flushing of the cache for the volume Deep1 scheduled daily at 03.55 AM

The incremental copy of the current tapes for the volumes that provide two copies of the data have been discussed previously in this guide. Please refer to 3.2 Integral Volumes Sets Configuration.

The flushing of the cache is the function provided by the software via the mmarc command that instructs the software to write to tape everything that is present in the cache. When users copy files to the various archives, the data is kept in the cache until the high watermark is reached. To ensure data protection and avoid that problems with the filesystem where the cache is residing could put in risk of data loss, every night the scheduler initiates the procedure that writes down everything that is in the cache that has not yet been written to a tape. These operations have also been configured to ignore the low watermark present on near1 and near2, thus flushing the cache completely of all the files pending on it and write those on tape. The option to achieve that is -e when using the mmarc command.

This has been scheduled at different times for each volume to make sure that the operations are completed using one drive only, in case other users are still trying to read data from the system. Moreover the scheduling has been configured so that every incremental copy of the tapes will occur once everything uploaded by the users during the day has been already written to it, thus effectively synchronizing the two tapes to the last update of data.

To retrieve the list of jobs scheduled inside the library you can use the command:

1 2016/01/29_09:03 ICOPY deep2 MltIncrCopy Waiting For Time 0: 2016/02/11_07:00 2016/02/11_07:01
2 2016/01/29_15:24 ICOPY near2 MltIncrCopy Waiting For Time 0: 2016/02/11_06:00 2016/02/11_06:01
3 2016/01/29_15:56 Cache Flush Deep 1-e Mmarc Waiting For Time 0:localhost 2016/02/10_23:55 2016/02/10_23:55
4 2016/01/29_15:57 Cache Flush Deep 2-e Mmarc Waiting For Time 0:localhost 2016/02/11_00:55 2016/02/11_00:55
5 2016/01/29_15:57 Cache Flush Near 1-e Mmarc Waiting For Time 0:localhost 2016/02/11_01:55 2016/02/11_01:55
6 2016/01/29_15:57 Cache Flush Near 2-e Mmarc Waiting For Time 0:localhost 2016/02/11_03:55 2016/02/11_03:55

which provides information on the scheduled jobs and the latest running times. More details can be requested using the option -l.

Edit | Attach | Print version | History: r3 < r2 < r1 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r1 - 12 Oct 2017 - StuartBorthwick1
This site is powered by the TWiki collaboration platform Powered by PerlCopyright 2008-2014 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.