Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
/ 2004-04-07 16:00:40 -0500
\ Todd Denniston:
> What about a slight modification to your procedure
> Lars get your clue stick, I bet you'll need to beat me with it :)
>
> Andreas Semt wrote:
> >
> <SNIP>
> > Okay, now I have two possible ways to go, but can I suggest a third way
> > (cause I won't use LVM2, so only one way left ...)?
> >
> > The situation is:
> >
> > nodeA (Heartbeat and drbd active), nodeB (Heartbeat and drbd standby).
> > To make a clean and consistent snapshot i will do following (all on nodeB!):
> a0) on Node A at 01:59 per cron { while [ time != 0201 ]; do sync;done }
you only raise your chances to get a clean view.
you can not *guarantee* a clean view.
> > a) stop heartbeat (at 02.00 pm per cron for example), i.e. that means
> > for heartbeat: "nodeB is dead!, i am the only one left!"
> > b) -> <not right>
> > c) shut down drbd (or heartbeat shut down drbd)
> Now drbd knows nothing about the other node.
> c1) violate a drbd tenet:
> mount -t ext2 -o ro /dev/device_under_drbd /mnt_point
> # perhaps a thinko here, you do not want the
> #journal ran because that would change the under device
> #with out drbd knowing. (assumption ext3 filesystem)
> c2) make backup from now snapped device
> c3) umount /mnt_point
> c4) skip to (e)
c1 to c4: nope.
replace with do "lvcreate snapshot", and start drbd again.
then, you can use the snapshot.
But I just do not understand why you insist on doing it this way around.
Use LVM2 on top of DRBD, and you don't need no hacks.
This way you can guarantee a clean snapshot (at least "clean" from the
file system point of view; to guarantee "cleanness" of the snapshot for
the data of the services running on top of it, you probably need some
additional scripting anyways...)
lge