Forum Post: Appserver problems: agent/broker communication issue or fault in OpenEdge 11 64bit

Status
Not open for further replies.
C

CABI

Guest
OE11.3 (64bit), Windows 2008 R2. 9 Databases, 11 appservers all "State-reset", fairly standard settings. Currently testing prior to upgrading. Case registered with Progress help 6 days ago. During testing I asked 8 users to access 3 applications that made use of 3 appservers. Afterwards when I use OEE or asbman scripts to close the appservers some remain in a "STOPPING" state (asbman... -query) with 1 agent available. During an earlier test one appserver would not allow more than two users to connect (initial servers= 5, min server= 2, max server = 50). After I killed it (asbman... -kill) it gerenated a "port in use" message when re-starting it (no server reboot). Other appservers that end up in a stopping state after a manual stop request and then killed will only restart if the Brokers' port number is changed (TCPview doesn't list the port) or the server is rebooted. It's not always the same appservers that are affected. It seem that where a Borkers port get used by an _mproserv process that a problem arises. I notice that during user testing a lot of _mproserv.exe processes are started some of which used the ports for some of the appservers, even for those that were not accessed during the testing. At the time of writing with no users on the systems I have 8 Appservers running (a total of 11 agents available) and TCPview shows 11 _proapsv.exe and 8 java.exe's procesess with ports matching those for the brokers and agents. So all as expected. I also have 9 databases running (with bi and ai) and assuming 3x _mproserv per database, I should expect 27_mproserv processes, but TCPview shows 61 _mproserv processes. The admin server, nameserver and appserver logs show no obvious error messages [not sure what to look for in the admin and nameserver logs]. Any ideas on what could be going on and what I should be looking for (progress and windows server setups). I can shift the Broker's port ranges upstream by a goodly chunk but I feel this just delay the inevitable perhaps as _mproservs occupy more and more ports?

Continue reading...
 
Status
Not open for further replies.
Top