Using cron jobs
You may use the standard Linux cron facility on the Vega time-sharing service to execute scheduled commands. This is not intended to be used to start long-running or very resource-intensive jobs - the High Performance Computing service is provided for that purpose. You are also requested not to schedule jobs to be run at very high frequency (e.g. more than once an hour). CIS reserves the right to kill any jobs that compromise the usability of the service for other users.
Cron jobs are defined using the standard Linux crontab command: refer to its manual page for details.
With effect from 20th October 2010, the cron service on Vega is clustered across all the Vega servers and backed up on the NetApp filers. This means that:
- Using the crontab command as a particular user from any one of the Vega servers manipulates a single crontab file for that user, shared by all the servers, so it doesn't matter which Vega server you are logged into when you do this.
- At any one time, only one of the Vega servers will be running cron jobs, on behalf of the whole Vega service.
- If a particular Vega server is down for maintenance, or because of a fault, users' cron jobs will still get run.
- All users' crontab files are regularly snapshotted and backed up, and so can be recovered, by request to the Service Desk (for instance, if accidentally deleted).
If you need to know which Vega server is currently running cron jobs (for instance, if you want to kill one of your running jobs), use the command
crontab -c from any Vega server: this will display the name of the current cron server, and then you can log into that server explicitly.
Please be aware that cron on the legacy CIS Unix service (the Solaris servers altair and deneb) is not clustered in any way, and CIS strongly recommends running cron jobs on Vega instead.