[DRBD-user] Migrating 0.7 to 0.8

Hildebrand, Nils, 232 Nils.Hildebrand at bamf.bund.de
Tue Sep 9 11:30:51 CEST 2008

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


Hi,

I joined this mailing-list since I ran into some problems when trying to
update from 0.7 to 0.8.

It seem that the IP-protocol used for drbd-communication has changed, so
that 0.7 can't talk to 0.8.

So I would like to find a migration path with minimal downtime for the
drbd-nodes...

My setup is:

Two heartbeat-cluster-nodes running XEN. The XEN-DomUs reside von
drbd-devices (one per DomU), the drbd-device uses a local LV as backing
device. I use a local, separate LV for the drbd-disk (so I don't use
"internal", which makes things easier in this case).

I imagine the following upgrade-path:

0. Have all drbds on Node A in primary 

1. Move all DomUs from the drbd-device (0.7) to the local LVs

2. Upgrade the drbd-disk-devices from 0.7 to 0.8

3. Reattach the LVs to drbd 0.8

4. Repeat 2 + 3 on the "secondary" side

5. Connect primary and secondary side


How can I do this with minimal resync time?

between 1 to 3 the drbd-disk-device on the primary side will not be able
to track block changes on the lv - so I will propably have to use the
new "verify"-mechanism before connecting the network as well - without
loosing the existing data on the lv (where does drbd store the
block-checksums?).

So I hope someone out there has done this already and can give some
advice...


Kind regards

Nils



More information about the drbd-user mailing list