[Tutor] schedulers

[David Rock]

> On Feb 28, 2019, at 17:23, Alan Gauld via Tutor <tutor at python.org> wrote:
> 
> On 28/02/2019 14:45, nathan tech wrote:
> 
>> but I'm thinking, should I do this?

Honestly, scheduling is the last thing to figure out.  As mentioned, it’s probably better to focus on backups and leave the scheduling to the OS tools that already exist. You need to write a solid backup routine, and an even more solid _restore_ routine.  Ultimately, the goal of backup software is to recover data.

Out of curiosity, what’s the driver behind doing this project in the first place?  
Do existing backup tools not do what you need (FOSS or otherwise)? 
Is this an academic exercise (i.e., for fun because you are curious)?
How complicated do you expect to get with the implementation?

If you are going to play with the scheduler, every 5 seconds is really overkill.  The reality is checking once a minute (even once every 5 minutes) is more than sufficient for even the most aggressive environments.  If you really need sub-minute starts, this is probably not something for a casual project.

have you thought about backup windows?  Defining the start time is one thing, but will you have an end time?  How long of a delay before you don’t try to start anymore?  What about media resources? How long will you wait if you have tape devices all in use?  Is this just a disk-based backup system?

Good backup software is not trivial to write.  Maybe if we better understood your goals, we could better direct you.


— 
David Rock
david at graniteweb.com