I have gotten my two semi-identical systems running with mirrored zpools and I have been doing some sending and receiving without incident. I'd like to get some feedback on my simplistic approach to syncing the two. Here's what I've been doing that's working on my two systems - odin and fenris:
on odin (the source):
which results in a nice baseline copy of odin on fenris.
again on odin as time progresses:
which results in an up to date copy of odin on fenris using an incremental diff from baseline.
on both:
This seems to be working swell as the incremental is pretty small after a week and I'm guessing so long as I haven't done any massive maintenance on the source, should be fast every time it runs.
Does this seem reasonable? I figure if I wanted to, I could re-baseline every so often or... could I rename a snap-mod and use that as the base for a diff?
- will
on odin (the source):
Code:
zfs snapshot -r prime@snap
zfs send -R ... | pv | ssh fenris zfs recv...which results in a nice baseline copy of odin on fenris.
again on odin as time progresses:
Code:
zfs snapshot -r prime@snap-mod
zfs send -vi ... | pv | ssh fenris zfs recv... (for each vdev)which results in an up to date copy of odin on fenris using an incremental diff from baseline.
on both:
Code:
zfs destroy -r prime@snap-modThis seems to be working swell as the incremental is pretty small after a week and I'm guessing so long as I haven't done any massive maintenance on the source, should be fast every time it runs.
Does this seem reasonable? I figure if I wanted to, I could re-baseline every so often or... could I rename a snap-mod and use that as the base for a diff?
- will