at times when I have to restart or say stop the service it starts re-sync which takes about 1 hour on a gigabit crossover. so that brings down my uptime by 1 hour.
drbd only really seems useful for (cheap) HA systems, so the downtime of a slave shouldn't kill you. It's the uptime of the cluster that's important, isn't it?
and it syncs block by block instead of incremental. thats the problem.
That would be consistant with the way RAID systems works, and drbd is basically just RAID over the network.
If you want something that doesn't sync that way, you probably want shared storage. That means either a SAN, or a shared SCSI array.
-- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://www.redhat.com/mailman/listinfo/redhat-list
