[Linux-HA] Cant failover drbd fs when fs is busy

[Alan Robertson]

Yan Fitterer wrote:
> Christophe Zwecker wrote:
>> Alan Robertson wrote:
>>
>>>> Well anyway mysql and fs didnt start on node2 , i suppose because the
>>>> fs couldnt be unmounted because of the copyjob.
>>>
>>> Depending on what your policy was, that could cause it to be marked as
>>> unmanaged (IIRC), or it could cause the machine to reboot.
>>
>> i dont ever want any ressource to be in state unmanaged, how can I set
>> the machine to reboot instead ? do I need stonith for this ?
> 
> Yes
> Only alternative (not always possible...) is to improve the RA involved
> to ensure it *always* manages to successfully stop a resource, no matter
> how broken or hung it is.
> 
> This will not help you deal with failures where a node stops responding
> to others though... That can *only* be addressed by fencing (i.e. STONITH)
> 
>> By using the gui i havent seen an option to reboot when i look at
>> ressource operations...
> 
> You'll need to implement a STONITH resource, according to the type of
> STONITH device you use.

Although for this case, he'll also need to implement a "suicide" STONITH
device - because the node is otherwise sane, it can reboot itself.


-- 
    Alan Robertson <alanr at unix.sh>

"Openness is the foundation and preservative of friendship...  Let me
claim from you at all times your undisguised opinions." - William
Wilberforce