IUBio

ACEDB 4.1 Client/Server

Staffan Bergh staffan at biochem.kth.se
Fri Jun 14 10:04:56 EST 1996


In article <4pr9he$hee at mserv1.dl.ac.uk> Dr Mike Arnold 
	<mike at nasc.life.nottingham.ac.uk> writes:

[some stuff deleted]

>Also, multiple client connections are forked from one session. For
>example, if I query classes and then click a returned link, another
>client connection is established - this eats into CPU and is probably
>not what was intended! 

Excuse me if this is really basic ...

If you make sure to do a closeServer call at the end of each 'transaction' this
problem will go away - in fact I think the openServer and closeServer calls in
Aceclient are a bit misnamed: they actually pertain to _clients_, not the
server. With the rpc setup, if you start a new client (with the openServer call)
then if no server is running a new server will be started (by inetd) *and* a new
client (although I've seen some problems with that bit - it seems the first
client just hangs ...). The following openServer calls will just connect to the
already running server. closeServer will close the active _client_, not the
server. The server will kill idle clients after a while (one of the timeouts in
the command line for the server tells how long a while) but if you have lots of
openServer calls in the script and don't close the clients as you go, it's easy
to overload the server - on my machine, when it gets up to a hundred active
clients or so, the server starts to return errors.

Cheers /staffan

Staffan Bergh
Biochemistry, KTH, S-100 44 Stockholm, Sweden

email: staffan at biochem.kth.se           + Don't let that horse eat that violin
phone: int+46 8 790 9230                +               cried Chagall's mother
fax: int+46 8 24 54 52                  + but he kept right on painting
                                        +             -- Lawrence Ferlinghetti

<A HREF="http://www.biochem.kth.se">Webmaster</A> and 
<A HREF="http://www.biochem.kth.se/MycDB.html">MycDB maintainer</A>






More information about the Acedb mailing list

Send comments to us at biosci-help [At] net.bio.net