[DRBD-user] DRBD8 and GFS issues

Tiago Cruz tiagocruz at forumgdh.net
Thu Jun 12 17:17:26 CEST 2008


Florian and Gordan,

Many thanks for your reply, help me so much!

On Thu, 2008-06-12 at 13:09 +0100, drbd at bobich.net wrote:

> Therein lies your problem. You need a proper fencing device for GFS 
> operation to transparently continue. If you are using fence_manual,
you 
> have to run fence_ack_manual manually on the remaining node (see man
page 
> for details) to get it to allow access to GFS again.


I did this and so I got access again to FS, many thanks! Now I'm
studying one better way to fence my poor nodes (RSA/DRAC).

When I tried to put my node2 again on cluster (I've dropped all using
iptables), I got:

Jun 12 11:33:58 hotsite-bsb-la-2 openais[3062]: [MAIN ] Killing node
drdb_hotsite-1 because it has rejoined the cluster with existing state 
Jun 12 11:33:58 hotsite-bsb-la-2 openais[3062]: [CMAN ] cman killed by
node 1 because we rejoined the cluster without a full restart

OK. I've tried to restart and lost the machine:

        echo o > /proc/sysrq-trigger ; halt -f

Jun 12 11:35:14 hotsite-bsb-la-1 kernel: drbd0:  old =
{ cs:NetworkFailure st:Primary/Unknown ds:UpToDate/DUnknown s--- }
Jun 12 11:35:14 hotsite-bsb-la-1 kernel: drbd0:  new = { cs:Unconnected
st:Primary/Unknown ds:UpToDate/DUnknown s--- }
Jun 12 11:35:14 hotsite-bsb-la-1 kernel: drbd0: conn( NetworkFailure ->
Unconnected ) 
Jun 12 11:35:14 hotsite-bsb-la-1 kernel: drbd0: receiver terminated
Jun 12 11:35:14 hotsite-bsb-la-1 kernel: drbd0: receiver (re)started
Jun 12 11:35:14 hotsite-bsb-la-1 kernel: drbd0: Considering state change
from bad state. Error would be: 'Refusing to be Primary while peer is
not outdated'
Jun 12 11:35:14 hotsite-bsb-la-1 kernel: drbd0:  old = { cs:Unconnected
st:Primary/Unknown ds:UpToDate/DUnknown s--- }
Jun 12 11:35:14 hotsite-bsb-la-1 kernel: drbd0:  new = { cs:WFConnection
st:Primary/Unknown ds:UpToDate/DUnknown s--- }
Jun 12 11:35:14 hotsite-bsb-la-1 kernel: drbd0: conn( Unconnected ->
WFConnection 

So... how can I prevent this? "Refusing to be Primary while peer is not
outdated".
I wanna to mount/use my FS when the second peer its crashed and your
status saw unknown :)
Is it possible?

Many thanks again!

-- 
Tiago Cruz
http://everlinux.com
Linux User #282636






More information about the drbd-user mailing list