We have a customer that has a limit of ten oengines set to handle all incoming web/app requests.

Generally speaking this is more than sufficient. At times however, when all ten oengines are active and the cpu is lingering in the high 90's, I have scrolled through the task manager and found 100 or more instances of the oecgi running.

As a novice, I'd guess that these are representative of the number of actual requests coming through and therefore it's no wonder that all ten oengines are busily chewing up the cpu. But, that's an assumption on my part so I'm asking if this is expected behaviour or could such a large number of oecgi instances be indicative of something going awry?

When I notice this happening it seems to be from two issues:

One, each click by the web user generates a oecgi hit. So, are they getting impatient and clicking frequently? Do you have any "Please wait" or spinning gif to advise them to wait?

Try running this in the java alternative to oengineserver service. Watch it in debug mode 2. It could be a system notice to the desktop (unwanted occurence), or some reason a debug is being thrown.

Dave Harmacek - Harmacek Database Systems - near Boston, MA USA

Yes - Dave probably has identified the "leaning" as the culprit.

there are still some ideas i have for bryan about the spinning "please wait" but i am not going down that path until oi 10.1 is out.

BUT

in the mean time.

when a request comes in lock that session until you have draw or redrawn or partially drawn your page.

then before your "return" unlock that session.

this way MOST but not all of the "Leaning" gets thrown away.

the rest of the "leaning" is how the "please wait" can be changed

Bri

Thanks gentlemen.

You have confirmed my suspicions and clarified my understanding.

In this scenario, we provide API's but are not directly involved in the front end development so there's not a lot I can do with regards to locking and redrawing. That's ok, at least I now better understand what I'm seeing.

I have seen end users set up in such a way that just the loading of one page would send a hundred or more requests so if each of those results in an instance of oecgi in the task manager, these apparent extremes make more sense.

Don't get me wrong. It makes no sense that a user would be set up that way but they sometimes are and if they were to log in, then what I'm seeing makes more sense.

Thanks for the help.

Mark I, SEE.

I'm wondering if your API's are using RESTFul?

Dave Harmacek - Harmacek Database Systems - near Boston, MA USA

You have wondered correctly

View this thread on the Works forum...