Java EngineServer Overlapping Recycling? (OpenInsight 64-bit)
At 19 JUL 2022 08:25:06PM Joshua Goddard wrote:
Hi
You can configure IIS to start a new app pool before closing the existing app pool? I was wondering if the Java EngineServer had similar functionality?
In our setup, the MaxUptime is 6 hours (perhaps we could extend it to infinity hours thereby avoiding this issue altogether?). Is there a setting to make the EngineServer start a new engine before closing the existing one?
At 20 JUL 2022 08:37AM bshumsky wrote:
Hi
You can configure IIS to start a new app pool before closing the existing app pool? I was wondering if the Java EngineServer had similar functionality?
In our setup, the MaxUptime is 6 hours (perhaps we could extend it to infinity hours thereby avoiding this issue altogether?). Is there a setting to make the EngineServer start a new engine before closing the existing one?
Hi, Joshua. No, I don't believe there's a setting for that behavior.
You'll want to keep the MaxUptime because I think I recall in your environment sometimes oengines that run "forever" exhaust their resources somehow. MaxUptime ensures that they WILL close and a new, "fresh" engine will start.
The Engine Server is designed to only start an oengine when needed, so after an oengine/all the oengines are closed, if a request comes in that can't be handled by any already-opened-and-available oengine, a new one will start up. The _intent_ is that the startup process, while longer than just pulling an oengine out of the queue, won't be SO long that it's an issue. Are you seeing a problem where starting the oengine is taking "too long" (and for what definition of "too long")?
Thanks,
- Bryan Shumsky
At 21 JUL 2022 09:34PM Joshua Goddard wrote:
I haven't exactly timed it, but I would guess it takes about 5 seconds to start up.
What I do know for sure though is that whenever the engines stop and start, we get a spate of requests that take 10 seconds+.
Anyway, we are going to increase the uptime to 12 hours, and perhaps infinity hours, as the resource leak appears to no longer exist. We could also create a task scheduled jobs that restarts the engines at 03:00:00 am every day (outside of normal business hours)
At 21 JUL 2022 10:07PM Donald Bakke wrote:
We could also create a task scheduled jobs that restarts the engines at 03:00:00 am every day (outside of normal business hours)
We have clients that are configured to restart engines as well as OI-based unattended processes for this very reason.