Re: Information Required on Problems Occuring with Oracle Webserver on Solaris

The only help I can give you is to tell you that you're not the only one experimenting the problem you described...

We have exactly the same system setup you have and 2 ou 3 times a day it gets heavily loaded : up to 40 wrb processes can be running at the same time!!

Of course, we used the quick-and-dirty-and-immediate solution, which is to kill all the wrb processes, stop and then restart the Web servers. But I would like to hear about a permanent solution. Note that the OWS 2.1.1 doesn't seem to lose control about the wrb processes. Yesterday, for example, there was 33 wrb processes in the middle of the afternoon (with a load average of 36 !). We didn't send any signal to any oraweb21 or wrb21 process and, by the end of the afternoon, the load average was around 0.35 and there was only 4 wrb processes left.

I'm currently working on optimizing our Web PL-SQL procedures to trapp all errors and avoid useless SQL requests. We'll see next week if it's going to make any improvement about the behavior of our OWS 2.1.1.... But I keep in mind that the multiple wrb processes is an old problem for OWS (there was many bugs in version 2.0).

We also installed OWS 3.0 on a test machine but we haven't really tested it yet. That should come soon (I noticed that the wrb architecture seems to be very different for that version...).

If you hear about a solution for our common problem, I would appreciate that you let me know. (I'll tell you if I get the secret.)

sTEF

Colin Woods wrote:

> Dear All
>
> I need some information from people who have installed Oracle WebServer
> on Solaris.
>
> I currently have OWS v2.1.1 installed on Solaris 2.6, with Oracle v7.3
>
> Problems seem to occur in the webserver in that, the Web Request Broker
> child processes demand a large amount of CPU (e.g. I have record 6 WRB
> process running at once, 3 of which have child processes demanding 20%
> of the CPU each).
>
> As you can imagine this is 'killing' other important process that are
> running on our machine.
>
> Has anyone experienced the same or similar problems, and what solution
> did you come up with?
>
> OR
>
> If anyone has OWS v2.1.1 installed on Solaris 2.5.1, or OWS 3.0
> Installed on Solaris 2.6 also let me know of any problems you have
> experienced with regard to performance.
>
> My current solution is to either downgrade to Solaris 2.5.1(which is
> only millennium compliant with patches) or upgrade to OWS 3.0 (which I
> am told is more resource hungry) but I am not sure which to go for?
>
> Any information will be appreciated
>
> Colin Woods
Received on Fri Apr 24 1998 - 11:02:35 CDT