any advice on mirrored db and arbitrary cron

[cretaceous]

This is a "process" question...
I am building a site for a client who has some fixed ideas of how things should be, so I'd like any comments on the following:

1) he wants it backed up every hour
- I suggested it might work to do a single 24 hour backup and add to that a duplicate database so all updates/inserts etc are mirrored onto another server-
any advice?

2) there is a function where logged-in third parties can upload a text file for later processing - this is so they dont have to sit and wait while the processing happens

the process will be checking information on an external dataset, with multiple entries (perhaps a few hundred) which returns a fairly simple set of results - the results are shown to the user separately, with an emailed summary

it could possibly take a few minutes to process and it seems to me less trouble to do it at the time
is there a method to do something like this in the background?
otherwise I'll need to allow them to set pseudo arbitrary times for cron to trigger the process - which sounds a pita
(it would have to be be a fudged 'arbitrary' so maybe choose the hour only, then the cron checks if there is a file waiting for that time and if so triggers the process)

thanks for all and any advice !

[bpat1434]

I guess the easiest thing to do with the backups is to just tar up the mysql folder. For example, I have a database called "anp1review_db". In my Ubuntu server all my mysql stuff is held under /var/lib/mysql. So in /var/lib/mysql I have a folder called "anp1review_db". Inside that folder there are quite a few files. So if you tar up that folder, and then scp or ftp it over to your secondary server.

To update the database, you can use rsync to copy the database files from one server to another.

As for your other problem with text-files, I say the easiest way is to set a cron to be run ever X minutes where X is the estimated amount of time it will take that script to run (on average) plus an extra 5 or 10 minutes. That cron checks a directory on the server where there are text-files stored. It processes each text-file, deleting them as it goes (if it does everything successfully that is).

Hope that helps.

[cretaceous]

thanks for comments on the cron bpat

I don't have problems backing up the database - it just seems crazy to do every hour - so a mirroring of all inserts/updates onto another server would seem to be a solution - or I guess I can get the web host to set something up - at great expense probably but that's not my problem !