|
Hello,
> You're the one with the servlets, aren't you?
Yes.
> Also it doesn't have active or passive state, it only has an
> expiration timer.
I know. I just wanted to point out the conjunction of template
and server. Kind of redundant information, I guess :-)
> > The better (or a safer) method in my case seems to be to not only
> > count
> > connections but also take a look for existing
> persistance-templates and
> > possibly pick another real-server to avoid the scenery above.
>
> It might help, yes. I'm actually doing some test conducts an
> persistent
> based counters to use in conjunction with RS thresholds.
Same as I do. But I need to do more tests. I thought, the (indirect)
trick could be, to set lower-threshold to a value closely to the
minimum needs (one connection), to prioritise all empty servers.
> Another possibility might be to use the feedbackd framework to
> dynamically adjust the RS weights.
Will check.
Btw: for testing purposes ist would be fine, to be able to zero all
persistence-templates.
> Horms' reply to your inquiry ;). However I should like you to have a
> look at a semester work proposal I recently did and tell me if this
> would solve your problem:
"Implementation of a persistency-based counter for all available
load balancing schedulers. [...] This persistent session counter is to
be understood as addition to the existing active/passive connection
means
which is based on TCP state transition information" [1]
[1] http://pubwww.fhzh.ch/~rnibali/FACHSTUDIUM/SA/Semester_Work_LVS.pdf
Sounds perfect to me!
Grüße in die Schweiz!
Holger Kettler
--
HIS GmbH Hannover Holger Kettler
kettler at his de 0xCBBE85FB
|
