Re: [SSI-devel] DRBD Root failover again

[Kilian C. <kil...@st...>]

Hi again,

I have an other question about failover. To test it, I use kdb to=20
artificially hang the master node. Then, I wait for the secondary node to=20
take over and to become the root node, then I revive the previous root node=
=20
by typing 'go' at kdb prompt.=20

At this state, on the new root node, cluster -v shows that the previous roo=
t=20
is still in KCLEANUP state. So the first node can't join the cluster again,=
=20
since the nodedown process does not end.
And after a while, on the new root, I have a Kernel Panic saying that a=20
timeout occured waiting for nodedown to complete.

Any idea?

=2D-=20
Kilian CAVALOTTI                               Ing=E9nieur Syst=E8mes & R=
=E9seaux
Laboratoire STIX                                        =C9cole Polytechniq=
ue
=4691128 Palaiseau                                    Tel : +33 1 69 33 34 =
95