Setting up CTDB on OCFS2 and VMs ...

Martin Schwenke martin at meltin.net
Tue Dec 16 13:50:44 MST 2014


On Tue, 16 Dec 2014 20:02:54 +0000, Rowland Penny
<repenny241155 at gmail.com> wrote:

> On 16/12/14 19:05, ronnie sahlberg wrote:

> > I.e.  you should probably change your public addresses file and set
> > the netmask to 24

> So I changed the netmask to 24 and now neither of the nodes come up, 
> change back to 8 and node 1 becomes OK, back to to 24 and both UNHEALTHY 
> again.

So, what do you see if you run "ctdb scriptstatus" on an unhealthy
node?  This should give you the reason that it is unhealthy.  If it
says "monitor cycle never run" then please try "ctdb scriptstatus
startup".

I'm guessing that you've restarted CTDB on both nodes after making the
netmask change?   There's a small chance that you've used "ctdb
reloadips", which is missing some logic to deal with change of
netmasks... but it hasn't bitten anyone yet, to my knowledge, so is low
on the priority list.

> Now Steve said that this wasn't rocket science, he was wrong, it is!!

Not rocket science.  Just a bit harder than setting up Samba.  More
dependencies...

peace & happiness,
martin


More information about the samba-technical mailing list