[Deploying Sakai] Load balancer issues with heavy load

[Maurer, Christopher Wayne]

Whenever it had happened previously, I was pretty surprised to see it, as
I figured we had to stay on the same server the whole time too.
We are also beginning some investigation of terracotta!

Chris

On 9/1/11 1:24 PM, "Aaron Zeckoski" <azeckoski at unicon.net> wrote:

>I believe that Sakai pretty much requires sticky sessions unless you
>are using the terracotta stuff to share the session across lots of
>servers. Something may have changed here that I am unaware of but I am
>pretty positive this is still the case.
>
>-AZ
>
>
>On Thu, Sep 1, 2011 at 12:38 PM, Maurer, Christopher Wayne
><chmaurer at iupui.edu> wrote:
>> We have seen this periodically before, but never to the extent that we
>>have
>> for this start of semester.
>> What is happening is that folks start out on one server in our cluster
>>and
>> the load balancer decides that that node is too busy and sends them
>>over to
>> a different node.  For "normal" navigation activity, it's not really a
>>big
>> deal.  It does become a big deal when they are in the middle of taking
>>an
>> assessment, or posting an announcement or something like that.  As far
>>as
>> the user knows, they hit submit and they expect that their transaction
>> completed.  But it didn't, because they got sent to a new node.
>> Do other institutions see these types of issues?  Just today we finally
>> changed the setting to keep people locked to one server, so we're hoping
>> that helps in a positive way.
>> Thoughts from others?
>> Chris
>> _______________________________________________
>> production mailing list
>> production at collab.sakaiproject.org
>> http://collab.sakaiproject.org/mailman/listinfo/production
>>
>> TO UNSUBSCRIBE: send email to
>>production-unsubscribe at collab.sakaiproject.org
>> with a subject of "unsubscribe"
>>
>
>
>
>-- 
>Aaron Zeckoski - Software Architect - http://tinyurl.com/azprofile