Hi,
I really think that the GUI is confusing and what is not clear to me is the
real meaning of the "JDBC Database Login Default" and "JDBC Database
Connection Pool Defaults".
If you consider only one JDBC Sampler and configure everything in it OK, you
configure the pool dimension for this request, every other execution of this
request (other thread, other iteration of the same thread) will use the
pool.
If you use the defaults what do you want to do:
1. just provide a way (as other JMeter Defaults objects do) to offer default
values to Samplers' data fields, so Samplers can use it when the data is
missing at their level, or not use it if any data is specified instead. And
in the JDBC case it means that a Sampler would only inherit the connection
pool parameters, not the pool itself, and will use these parameters to setup
its own pool instance.
2. or would you consider the Defaults objects as a container for the pool
instance itself (what seems to be the present behaviour if all Samplers
using the same DataSource will share the same pool...)... But why is it
split into "JDBC Database Login Default" and "JDBC Database Connection Pool
Defaults" then ?
What I can imagine (JMeter more experienced users will correct me if I am
wrong or if it is in fact already the way things are working...) is that,
like any other Defaults object, the "JDBC Database Connection Pool Defaults"
should remain default values providers for Samplers (so case 1.).
But we would then need a kind of "JDBC connection pool manager" (a bit like
the Http Cookie Manager works) who would represent a connection pool
instance (case .2).
So Samplers hierarchicaly "under" the pool instance would have the choice to
use it (if nothing specified for DataSource, Pool size, ...) or make their
own connection (and provide themselves all the info needed).
Could be extended to other kind of connections, not only JDBC...
What do you think ?
Thanks,
Laurent
-----Message d'origine-----
De: Jeremy Arnold
A: JMeter Users List
Date: 19.08.03 23:09
Objet: Re: JDBC connection pool
Hello,
My experience with the JDBC sampler is rather limited, but I've been
looking at the connection pool code recently, so I can take a stab at
answering your questions. Anybody else can feel free to jump in and
claim I'm wrong. :)
From the code, it appears that there is really just one connection
pool per database URL. So if you have multiple JDBC samplers with the
same database URL, they will share a connection pool. I realize that
this probably isn't what you would expect from looking at the GUI.
I can't comment on the JDBC Database Connection Pool
Defaults...haven't looked at that.
I've been doing some work recently on rewriting the connection
pool. If you have some thoughts about how you would like it to work,
please feel free to send them along. This work is planned for the next
major version of JMeter, but we could consider putting it in a 1.9.x
release if there is demand.
Jeremy
http://xirr.com/~jeremy_a
Laurent MEDIONI wrote:
>Hi,
>does someone know what the connection pool for JDBC sampler exactly
means ?
>Is the pool available to all Loops of all Threads ? Only for this
Sampler ?
>
>And why are the 2 param "Number of Connections in Pool" and "Max Usage
For
>Each Connection" mandatory at the Sampler level ? What about using a
JDBC
>Database Connection Pool Defaults then, if everything has to be fixed
at the
>Sampler level ?
>
>Thanks a lot
>Laurent
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: [EMAIL PROTECTED]
>For additional commands, e-mail: [EMAIL PROTECTED]
>
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
