In our environment we have two DBAs to run some cron jobs defined in cron_jobs.txt file. Frequently we need to chan ge timing and commands in that cron_jobs.txt file and reload it.

Problem is, if DBA_1 edits and reloads that file ($ crontab cron_jobs.txt) it is loaded under his useraccount. Now, DBA_2 edits the same cron_jobs.txt file and reloads. The new sets of jobs will run under DBA_2 useraccounts. So, basically same jobs are running twice.

For an example, a job is set to run on 3PM daily by DBA_1. Few days later, DBA_2 changes it to 2PM and reloads. Now we have same job running at 2PM and 3PM.

Is there any way to merge the jobs? Liek no matter who loads the cron_job.txt file, only one set of jobs should be running? By the way (DBA_1 has admin access to UNIX, DBA_2 doesn't). We are using Sun Os 5.10. A solution will be appreciated.

Thank you.


5 Answers 5


Setup a separate account for this purpose, and let cron jobs run under that account only.

This is a best practice and offers many benefits beyond fixing your problem:

  • Nothing to worry about if either DBA is hit by a bus, or other
  • The real user account environments don't affect your production cron jobs
  • Coupled with sudo, it gives good logging/accountability

In my previous company, I disabled cron & at for all real user accounts.


Use a role account rather than tie things to a regular user account. Then allow both users to run this command.

sudo crontab -u dba_cron_user -e


One way to do this would be to have a semaphore file that both users can write to. Have the cron job check on the existence of this file and the date of the file - if it's today's date than the job has already been run - so no need to run - the 2nd attempt to run the file would exit. The problem with this approach though is that the earlier run would always win.


Normally on says,

crontab -e

and, edits the file. In which case the earlier edits are present at the time of making changes. This keep the crontab (of one user login) consistent.

Another way would be to fetch the current file, append it and then load it.

crontab -l > buffer.txt
cat new_command.txt >> buffer.txt
crontab buffer.txt

Here, the trick required will be to script the merge of new_command.txt into buffer.txt so that replication is eliminated. The human user would normally do that correctly when using crontab -e.


You could create a hard or soft link from the first user's crontab file(/var/spool/cron/DBA_1) to the second user's crontab file(/var/spool/cron/DBA_2). The downside to this method is that both users share the same crontab, and neither gets an individual file.

The above mentioned locations are correct for RHEL/CentOS; they may be different in another distribution.

  • Wouldn't this cause jobs to be run twice (once for each user)?
    – Keith
    Feb 16, 2017 at 17:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.