[Linux-cluster] Advice on Storage Hardware

[Michael Will]

There is no single partition limit, it was more that I said
we have done active/passive rather than active/active, where
all partitions belong to the currently active node rather than
having an active/active where some belong to one and the others
to the other.

Bottom line: what you are trying to do is certainly possible.

Michael

David Brieck Jr. wrote:
> On 11/11/05, Michael Will <mwill at penguincomputing.com> wrote:
>   
>> In our case there would be a serial and an ethernet heartbeat between
>> the systems.
>>
>> If one system detects the lack of responsiveness of the other system, it
>> shuts it down,
>> does an IP takeover and mounts the partition that belonged to the other
>> system and
>> now offers it under the new IP address. When the other system has been
>> brought up
>> again, and has been considered stable, it is being advised to take back
>> it's ressource,
>> which does initiate the other system to release it and once that has
>> been confirmed, it
>> takes over the IP address and starts it's nfs service for that partition
>> again.
>>
>> I have only configured this for active/passive setups with a single
>> partition so far, but
>> to extend that to two partitions and an active/active setup should not
>> be too hard.
>>
>> Michael
>>     
>
> Thanks Michael. We probably would need more than one partition,
> probably somewhere around 5, just so we keep data segregated and
> manageable. I still have yet to get my hands on a setup to test things
> out, but from all the docs I've been reading the number of partitions
> shouldn't be a problem.
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>   


-- 
Michael Will
Penguin Computing Corp.
Sales Engineer
415-954-2822
415-954-2899 fx
mwill at penguincomputing.com