John,
Try setting discovery.mode=-1 and ensure the other properties are set according to this[1].
Darren
The 'discovery.mode=1' setting sets the presto admin server to "no discovery,' which is the opposite to the passive/active discovery behavior that I'm attempting to enable.
So far, I have:
- 2 presto instances sharing a repository
- Disabled caching for both (rdsApplicationContext.xml)
- Copied the DB service code/config to common directories on both boxes, and set the global params to point to their abs paths.
- Followed the 5 steps listed under 'Configuring Cluster Discovery'
This returns the error I listed in my initial posting.
If I set 'discovery.mode=0' and set the node list to include both presto intances, the admin server will display both presto nodes successfully.
Just for clarity on the issue, did you also set up the configuration required for each Mashup Server? There is a link in #2 of the Configure Cluster Discovery topic that covers this configuration requirement. See:
http://www.jackbe.com/prestodocs/v2.7.0/prestoadmin/index.html?href=PrestoAdminLibrary.1.76.html
<!--Session data-->
Sara, technical writer/jackbe
If I've understood the question correctly, I've left the subagent.config files alone. The documentation states, 'The default configuration in this file (subagent.config) works with the default cluster configuration for the Presto Administration Server.
The entry in both the server and non-server presto node subagent.config files are: esd.serviceUrl=service:jmx:jdmk-http://localhost:8282
The node.list entry in adminconfig is: service:jmx:jdmk-http://localhost:8282,service:jmx:jdmk-http://10.20.5.50:8282
Let me know if I missed something.
I'm attempting to configure two instances of Presto to be for active discovery. Both already share the same user and application repository successfully.
After modifying the admin server's adminserver.config as described in the online doc and then restarting the server, I saw the following error:
INFO: Deploying web application archive adminserver.war
2010-07-22 18:02:29,298 ERROR [org.springframework.web.context.ContextLoader-205] - <Context initialization failed>
org.springframework.beans.factory.BeanDefinitionStoreException: Error registering bean with name 'prestoMasterAgent' defined in ServletContext resource [/WEB-INF/classes/applicationContext.xml]: Could not resolve placeholder 'node.list' at org.springframework.beans.factory.config.PropertyPlaceholderConfigurer.processProperties(PropertyPlaceholderConfigurer.java:242)
Was there a step I missed that caused the the prestoMasterAgent to attempt to resolve node.list? According to the doc, it should be commented out. Looking at the applicationContext.xml, it appears to be required?
If I uncomment this line: node.list=service:jmx:jdmk-http://localhost:8282 and restart the admin server, I no longer receive errors, but the admin console does not show any servers either.
Verified 8282 is reachable using telnet localhost 8282.
If I make a single line change, disovery.mode=0, and restart the server, the admin console will show localhost as a presto server.
I've attached the adminserver.config. Please let me know if I didn't follow some steps correctly.
Regs,
John E.