D
dbeavon
Guest
That is interesting that more CPU's causes an overall performance decrease. Have you turned up tracing (4GLTrace) and review the timing differences in both cases? Another interesting thing would be to set the max sessions per agent to "1" and see if multiple PASOE agent processes would perform as well as they did in "classic". With only *one* PASOE agent per session (_mproapsv), the result might be similar to when there was only ever one _proapsv per session in the past (in "classic"). There aren't that many resource conflicts between the independent sessions (threads) in a PASOE process (_mproapsv) but I know of a few (eg. xml, clr bridge). You may have introduced your own resource conflict via a custom or third-party component. Can you repro the problem when VMWare is removed from the equation? (IE. repro on a physical host with more than one core)? If VMWare is a necessary factor to cause the problem then Progress may not want to work on this. I suspect Progress OE is totally agnostic about whether you are running on a VM or a physical. The problem could just as well be sent to VMWare.
Continue reading...
Continue reading...