User Experience of Self Management

Attendees:
Sankara, Siva, Anissa, Jothir, Dhiru, Nazrul, Vince, others.

Summary:

• Overview
  • Slide 5-18 provides an overview of Self Management.
  • In GlassFish v2, we are adding new Cluster Events. Currently, this has start/stop. Need to resolve issues with missing GMS events.
• Predefined Rules
  • We will be providing Self Management rules out of the box in v2. This is very exciting!
    • Self tuning JDBC connection pool - Suggestion was made that we can potentially extend this for JMS and other pools.
    • Out of memory detection will send out an alert. Clean up is not well defined yet. Could potentially prune memory usage in the system when it reaches a certain thresh hold.
    • Thread/instance hang detection will send out an alert. Action could potentially restart the instance.
    • Disk full will send out an alert. Action may also clean up server logs.
    • Dead queue message alert will send out an alert when ever a message can not be delivered.
• Documentation
  • All new rules will be clearly documented in our docs.
  • Suggestion was made to include 1-2 line description of the rule so that user has a context of the rules.
  • We should add a link to the self management docs in Admin GUI.
  • There were some concerns about using the rules in production. Suggestion was made to provide FishEye links to the src in our docs. We also should build an GUI to show the audit messages so that user can find out the history of a management rule.

Open Issues:

• Cluster events only has start/stop. We should consider adding other GMS events such as server crash, add/remove of server instance to the cluster, etc. - Sankara
• Pool self tuning should be expanded to all pools (HTTP, ORB, EJB, etc.) - Sankara
• Provide additional info link from GUI - Anissa, Sankara
• Provide short description of the rules in GUI - Anissa, Sankara
• Address production usage concerns in next release - Sankara
• Blog about new rules when the are available - Sankara