Kbase 20873: WebSpeed & AppServer Remaining in Starting State
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
10/15/2008 |
|
Status: Unverified
GOAL:
What are some reasons the agent can remain in STARTING state?
FACT(s) (Environment):
Progress 9.x
WebSpeed 3.x
FIX:
1. The agents are waiting on a TCP port, on AIX this process was very slow if there were a lot of ports in use in the ?minport to -maxport range. Verify and start each broker with a specific range of agents/servers -min and -max ports.
2. The agents are trying to connect to a database. If the database is not up and running, you have to wait for the timeout to occur before the agent dies. Start the database first, before starting agents/servers.
3. There is a lot of CPU usage during starting up. Due to the lack of CPU resources, agents/servers times out before it can finish startup.
4. Java memory, increase two JVM parameters for $JVMARGS;
ms - memory allocation pool
mx - heap size
JVMARGS="-msXm -mxXm"
Each WebSpeed and AppServer broker(s) consume 5M of memory. Unlike WebSpeed agents, each AppServer server, is a java thread and consumes at least 1M of memory each. Add &/or change JVMARGS underneath each broker instance within the ubroker.properties file. Changing JVMARGS within java_env will change those values for all JAVA processes.
It's best to set the values in the ubroker.properties file so you can change each broker independently, without effecting all the other brokers. If you make changes in the ubroker.properties it should override settings in the java_env file.
More information can be found regarding JVMARGS and it's functionality within the following Progress Solutions: 19310, 19693 & 19707.
5. If upgraded to WebSpeed 3.1C on Windows platforms and are using the ubroker.properties file from the previous version, refer to Progress Solution # 21264 "WebSpeed 3.1C does not start with upgraded ubroker.properties"
6. If the problem is only occasional, and only some agents randomly remain in the STARTING status, and they are trying to connect to to remote databases, then there maybe routing problems on your network because of which some network packets are being lost.