[Linux-ha-dev] TNG_2004_ALPHA_2 with latest HEAD Fixes problems with 2 Node Power Down Failover Testing.

[Andrew Beekhof]

On Sep 17, 2004, at 7:56 PM, James Macon wrote:

>
> I have been running a 2 node test with a virtual ip address and Apache 
> and powering down nodes to see if they failover. I have a client that 
> connects to the virtual ip address and each node has an index.html 
> that specifies what node is currently serving the page. I had some 
> issues/problems with this running but with the latest code from Andrew 
> on HEAD, this test works repeatedly. I have been able to power down 
> and up to 8 times now and not one failure.
>

that is very cool!

A word of caution to people that might get carried away, TNG supports 
different types of failure handling (and even cooler, on a per-resource 
basis)... the one in operation here does not ensure that the old 
instance is fenced before starting the new one.  This could lead to 
data corruption and is definitely not a good idea for peoples' disk 
based resources :)

>
> Thanks Andrew for the support you have given me and I apologize if I 
> am stressing the code so much at this early stage.

Not at all, in fact I should be thanking you!  You've already helped 
pin-point a number of holes :)

>
> Next step is to verify this with 3 nodes.

That *should* work, I should also write a test case for it one of these 
days...

>
>
>  Jim Macon
>  Phone: (919) 543-7490  Tie Line 441-7490
>  Internet: macon at us.ibm.com
> _______________________________________________________
> Linux-HA-Dev: Linux-HA-Dev at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> Home Page: http://linux-ha.org/
>
-- 
Andrew Beekhof

"Ooo Ahhh, Glenn McRath" - TISM