[Pacemaker] Fixed! - Re: Problem with dual-PDU fencing node with redundant PSUs

[Digimer]

On 06/28/2013 10:39 AM, Lars Marowsky-Bree wrote:
> On 2013-06-28T10:27:54, Digimer <lists at alteeve.ca> wrote:
> 
>>> Basically, unless we can do this better, having multiple devices per
>>> fence topology level needs to be considered broken and might be better
>>> removed.
>> NO NO NO NO.
>>
>> Please do not remove this. I can not use pacemaker unless I can keep the
>> power rails redundant. What we have now may not be elegant, but it
>> works. I would be a very sad panda if this functionality was removed.
> 
> It was a bit tongue-in-cheek. But the syntax is so horrible that forcing
> it on users *is* broken; requiring this kludge when *all* fencing
> topologies specifying multiple devices per level will need this is not
> good and a support nightmare.
> 
> Also, it is *not* working fine for you, or is it? What happens when one
> of your fence devices cannot be turned on again?
> 
> Regards,
>     Lars

I would obviously like it to be cleaner, but I can live with it.

Yes, a failed "on" action would then fail the method. This is
sub-optimal as FenceAgentAPI says that only the "off" portion of
"reboot" needs to succeed. However, I don't consider this a show stopper
because "on" action of PDUs simply means "re-energize the outlet". If
the node blew up, it won't boot, but the "on" will still succeed, so the
overall method would succeed.

For this "off off on on" action to fail in a way that effects the fence
action, IPMI will also have to have failed and then we're into multiple
simultaneous failures which these clusters aren't expected to survive.

Again though, with all this said, I will be happy with just keeping this
existing functionality. It works. If Andrew decides that he doesn't want
to expand out paired "reboot" methods into distinct 'off off check on
on', I can still write a fence agent wrapper that does this. Then people
who want to keep pacemaker's config simple will have the option of using
these wrapper agents.

In fact, I'll write 'fence_apc_multi' as a proof of concept. Give me the
weekend to do this.

-- 
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?