Kbase P99927: WebSpeed agents are no longer associated with broker and become a defunct or run away process
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
9/17/2008 |
|
Status: Verified
SYMPTOM(s):
WebSpeed agents are getting out of control of broker
WebSpeed agents are no longer associated to broker
Agents are no longer seen by the WebSpeed broker
Agents become defunct or run away processes
wtbman -q or WebSpeed status page from Progress explorer no longer displays the agents PID
java.lang.NullPointerException
at com.progress.ubroker.broker.ubWSserverThread.parse(ubWSserverThread.java:371)
at com.progress.ubroker.broker.ubWSserverThread.processServerStateChg(ubWSserverThread.java:329)
at com.progress.ubroker.broker.ubServerThread.processEvent(ubServerThread.java:1094)
at com.progress.ubroker.broker.ubServerThread.mainline(ubServerThread.java:436)
at com.progress.ubroker.broker.ubServerThread.run(ubServerThread.java:330)
Above errors are seen in broker log file
WebSpeed broker is able to continue spawning new agents, even though some agent processes become defunct or run away
collectStatsData=1 is set within ubroker.properties file for the broker
Fathom is installed on the server (with WebSpeed) but is NOT running
Disabling JIT does not resolve the problem
FACT(s) (Environment):
WebSpeed 3.1D
WebSpeed 3.1E
OpenEdge 10.0x
OpenEdge 10.1A
All Supported Operating Systems
CAUSE:
Bug# OE00126050
CAUSE:
The underlying problem here is that when we set "collectStatsData=1", both the AppServer and WebSpeed agents collect statistical data that is then transferred to the broker (the broker, in turn, gives this info to Fathom). The problem occurs during the update of shared data structures in the broker. The errors happened when multiple agents (App Server or WebSpeed) sent their statistical data and attempted to update these structures simultaneously.
FIX:
Option 1:
Set collectStatsData=0 underneath the broker section.
Option 2:
Upgrade to 10.0B05, 9.1E03 or 10.1A01 and later.