Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Rafael, let's back up a bit here.
On 11/28/2008 07:14 PM, devlists at soscpd.net wrote:
> Hello List
>
> I'm using a drbd Version: 8.0.12 (api:86) on gentoo kernel 2.6 to sync 2
> nodes. Last week, after a reboot, one node get disconnected and no more
> sync since. Last night, after other reboot, both nodes get connected, and
> the older (1 week) get the primary state, propagating the old data to the
> second node, messing up with all data. My question: Is there a way to roll
> back the data from the first node?
> --discard-my-data was tested with no success.
If I understand your situation correctly,
- first one of your nodes (let's call it "joe") lost the connection to
its peer ("jane"), and continued in disconnected mode for several days.
- then a few days later, joe and jane reconnected.
- now, without any further intervention, jane became primary, became a
synchronization source, and warped the entire cluster back in time
several days.
If that was really what happened, then something would be terribly
broken inside DRBD. That's absolutely not supposed to happen and DRBD
has quite a few nifty safety nets designed for preventing this type of
time warp.
So, to be honest, I find it hard to believe that that was really the
sequence of events. Can you please provide
- your drbd.conf;
- relevant log snippets from the "time warp" event;
- a detailed description of the administrative steps you took in the
process, compiled preferably with the help of your bash history file?
Thanks.
Cheers,
Florian
--
: Florian G. Haas
: LINBIT Information Technologies GmbH
: Vivenotgasse 48, A-1120 Vienna, Austria
When replying, there is no need to CC my personal address.
I monitor the list on a daily basis. Thank you.
LINBIT® and DRBD® are registered trademarks of LINBIT.