[Seaside] my site is completely dead..

[Sven Van Caekenberghe]

On 30 Jan 2013, at 13:54, Francois Stephany <tulipe.moutarde at gmail.com> wrote:

> On 30/01/13 10:15, Sven Van Caekenberghe wrote:
>> 
>> On 30 Jan 2013, at 10:06, Francois Stephany <tulipe.moutarde at gmail.com> wrote:
>> 
>>> On 30/01/13 09:33, Sven Van Caekenberghe wrote:
>>>> - most of the time, I have multiple images under a load balancer, statefull or stateless, to improve availability and capacity
>>> 
>>> Stateful = session affinity?
>> 
>> Yes. With Seaside you have to, with pure Zinc you can but it is not the default.
> 
> Yep, I guess that's another issue for Cloud deployment à la Heroku...
> 
>> If you reload http://zn.stfx.eu/status a couple of times, you will see the port changing (which means another member of the cluster is handling the request). If you reload http://zn.stfx.eu/session you will stay on the same cluster member (notice the session ID ending with a Route ID).
>> 
>>> Do you start images on-demand depending of the load or you let X images running all the time?
>> 
>> Right now, I do it statically: I just run 4 of them all the time.
>> 
>> One vm+image uses maybe 100 Mb, with current servers you get multiple GBs of RAM, that you can use for running multiple images, a db, memcached instances or other stuff.
> 
> Thanks !
> I know it's not really relevant without context but how much load can one image handle? It's just to have an order of magnitude…

You would have to define 'load' and even then, it really depends too much on the application (architecture) and/or frameworks used. The only way to find out it to try ;-)

Sven