Ok. Indeed, I never considered an interval which is not a common divisor of 60. Just out of curiosity I tried the crontab on my system and it runs at 00 and at 59 minutes of every hour! It looks like cron starts the count at 00 minutes in any case and applies as many intervals as they fit in one hour. Does it work like this on your system? Thanks.

Yes!

Besides, those who think

would execute the command every 30 seconds, don't seem to have tried it really. Because, when sleep 30 would cause the next command to be delayed for 30 seconds! So, there would be delay of 90 seconds!

So, logically, I have not yet figured out how to shcedule a job to be executed every 30 seconds.

My bash is a little rusty, so here is one method:

Write up a script/function to do what you want, then execute said script in the background, store the PID to a variable, then run sleep. At the end of of sleep run wait PID to see if the process finished. If it has not it will wait until that script has executed before continuing. You could also use pgrep to check to see if the process exist anymore, which would allow you to simply loop over the sleep command in a way such as (again, rusty from memory bash here):

That is just an example of course, but you should get the idea. That should prevent the process running several times at once while timing the process to run every 60 seconds. You could do something similar with cron, simply checking for the process before executing it via pgrep or some other method.

EDIT:

Wouldn't
<time fields> sleep 30; command
execute sleep 30; command every second, simply delaying the execution of the command but actually running the line every second?

Well, I think, when I checked it it was executing every minute instead.