[admin (no pm's please)]

Since Sunday is still hitting the slarti server, I have made some more changes.

1) The main backup is going to start earlier, but at a lower priority.
2) A pointless incremental backup will no longer occur.
3) Ford and Zaphod servers will take more of a role to take more backup load off slarti.

I hope this will get Sunday morning back to running smoothly. Slarti really has a lot of power and should not have issues at all.

[admin (no pm's please)]

We are now in the final phase of the backup.

Slarti seems to me to have run fine throughout.

Coincidentally Ford was hit by a nasty search flood (counter-measures now in place on the site effected) and had a few minutes of database connection issues.

I'm looking at tweaking things a bit more all the same.

[admin (no pm's please)]

Last Sunday due mostly to the clocks going forward there was a brief issue again.

But enough is enough on this, and the regime is now being totally changed.

Previously backups were pushed from a server to another server in the same datacenter, then pulled from there by a server in another data center, and finally copied again to an offsite backup system. All systems involved having RAID setups that are meant to survive a single disk failure.

So essentially the file system was preserved:

1) By RAID on the system itself.
2) A backup on the same system.
3) A backup in the same datacenter.
4) An offsite backup.
5) a further offsite to the offsite backup.

A much as I believe in backups, given the problems involved in (3) as it is when the push of that backup occurs that things go slow, I am eliminating this step.