Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Actually bringing the device in service on the standy side can be done
by with drbdmeta (see other e-Mail in this thread), however when the
former primary comes back (".. wow we could recover the machine and it
did not fully burn down :) .. ") it's difficult to solve the split brain
situation. A simple "drbdadm connect all -- --discard-my-data" does not
work or does not trigger the sync.
What needs to be (officially) done to safely trigger a sync (and control
the direction) after the described problem below ?
Any feedback appreshiated,
Robert
Robert schrieb:
> How can I easily set a outdated disk back to UpToDate ?
>
> Lets assume the situation where your primary data center burns down.
> First the network switch for the replication link burns and then the
> server itself. So first the standby disk is outdated by dopd. 2
> minutes later the primary server dies and heartbeat ties to activate
> ressources - as intended(!). How your boss comes and says - primary
> data center burned down ? what the heck, we have a cluster start all
> engines in the other data center. Your data (some minutes old but
> outdated) in the second data center is ALL you have, so you have to
> remove the outdated flag so start the ressource.
>
> How can someone remove the outdated flag ? drbdmeta ?
>
> Robert