Welcome to the SRP Forum! Please refer to the SRP Forum FAQ post if you have any questions regarding how the forum works.
Orphaned OEngines
Hola all,
My client has an issue with some of their Web-based sessions, which use OE, abending. When this happens, it seems there is still a session running on the LH Server. Is there a function or utility that can programmatically close down any orphaned OE sessions on the UD/LH service?
This part is all new to me, so thanks for the support!
Michael
My client has an issue with some of their Web-based sessions, which use OE, abending. When this happens, it seems there is still a session running on the LH Server. Is there a function or utility that can programmatically close down any orphaned OE sessions on the UD/LH service?
This part is all new to me, so thanks for the support!
Michael
Comments
All the extra OEngine occurrences are severely impeding performance.
Thanks again!
Michael
Some of those posts might have been mine as I have certainly seen that issue.
There are some eServer settings that should influence that behaviour:
(Ignore my actual values)
!Engine parameters //Number of Simultaneous connections to engine.0 = Unlimited. MaxConnections=0 //Number of engines allowed to start up. Default 10. 0 = Unlimited. MaxEngines=3 //Max time in minutes allowed active in memory. Default 0. 0 = Unlimited. MaxUpTime=0 //Max time in minutes allowed run a request before 'hung' and terminated. Default 0. 0 = Unlimited. MaxRunTime=3 //Time in minutes can be idle. Default 15. IdleTimeout=2 //Time in seconds between checks for idle and Uptime/Runtime. Default 60. IdleCheck=60 //Maximum number of Engines to close at one time (till next check). Designed to stagger on/off. Default 0. 0 = Unlimited. MaxNumToClose=0 //Keep alive as zombie processes (possibly reattached). Only for Telnet sessions. 0 will terminate. Default 0. KeepAlive=0
However, despite tweaking these to be super aggressive, I still had a site that would quickly have more than a dozen engines running despite a small MaxEngines number. This quickly ate up all thei Rev licences.
Unfortunately, I never got to the bottom of it. They were growing impatient so we implemented a Windows Scheduled Task that ran every so often (every 15 or 30 minutes) that called a Powershell script that killed off any OEngine that had been up for a 'long' period of time.
Not a great solution but they stoppped complaiaing so we had to take the 'win' and move on....
We have a lot of sessions that run from remote desktops, in addition to whatever is running on the users' desktops.