OK, I can't take this anymore. :)
> Now, I've been running for not even 1 hour with the skip-errors
> enabled.... A quick check, on ONE table...
> So there's already data missing on the slave.... *shrugs*
Yep, you told it to SKIP ERRORS. That means that if any query generates
an error, for any reason, it's just going to skip over it and move on.
So data was missing on your slave the moment you restarted replication
-- it skipped the UPDATE it had a problem with. From that moment on,
your replication was out of sync.
> And this is not the first time it's happening either... It really
> makes me doubt whether MySQL is the right approach to take to this
> whole replication vs data redundancy scenario.
You do realize that MySQL 5.x is ALPHA, right? Alpha doesn't mean
"works beautifully on production systems", it means "probably, very
likely, broken in some way". Why are you using an ALPHA version and
expecting it to work perfectly, and when it doesn't, "doubting" MySQL
and replication? If you want it to be stable, use a STABLE version.
Technical Yahoo - MySQL (Database) Geek