[dm-devel] Multipath and HSG80 phase 2

[Nicola Ranaldo]

> I'm going to continue in testing overall the package and giving you 
> feedback.

Ok... this seems to be a good morning!
I have now a veeeeeery (!) strange multipath environment up and running.
I report this hoping it will be a good starting point  in order to 
understand the final steps to fix my environment...

I rebooted the system and let the multipathd do all the stuff in order to 
create dm table, devices  and so on.
When all was done I killed multipathd and run the simple following batch in 
order to see wath's happening:

while true
do
multipath -l -v3
sleep 10
done

Restarting alternatively the primary and the secondary controller does not 
break the process writing on the multipath unit :) (done this operation 
about 30 times)
So i think the simple batch reenables failed paths well.

While running multipathd, restarting the controller all paths fail (i think 
it is normal) and multipathd seems to sleep until the controller wakes up 
again, but in this situation results are unpredictable, some time it resists 
only to a uniqe restart, some times gives directly write error opration on 
filesystem...
What the next step?

Another question, when failover does its work, while having no writing 
process disruption i have lost page write dm-0 kernel log.... is it normal?

Thank you

    Nicola Ranaldo