• Bug
  • Status: Open
  • 1 Critical
  • Resolution:
  • Terracotta Developer Console
  • kkannaiy
  • Reporter: sdcode
  • December 03, 2010
  • 0
  • Watchers: 1
  • October 11, 2011

Description

I’m running Terracotta found in Sourceforge ehcache-2.3.0 as a server array consisting of 3 nodes (Server1, Server2 and Server3) that are each Amazon EC2 instances. I’ve configured each instance to set its address as one of the specified addresses given in the shared tc-config.xml file. I have connected the Terracotta Developer Console v3.4.0 to one of the nodes and then I’ve been turning the nodes on and off to see how the nodes switch between being unreachable to passive to active.

Based on the terracotta/server-logs the nodes appear to be operating as expected. The problem is that the developer console seems to be unresponsive (in an intermittent way). For example, assume the developer console is connected to Server1 (active mode) and I take Server2 and Server3 down. When I restart Server2 and Server3, they will both successfully connect to the cluster (based on the logs of all three servers) but the developer console will show Server2 as red (disconnected) while Server3 is shown as blue (passive standby).

Initially I thought the devconsole was not receiving the updates from the server array at all but based on the devconsole’s log file, it appears that it is aware of the server array changes (at least vicariously through the corresponding server) but just not updating the icon colors appropriate. In general if I shut down a server, the corresponding icon color will go red but if I start a server the icon color is not updated to switch to blue (passive) or green (active).

While the icon colors are basically cosmetic, this led to serious confusion on my part as I was initially led to believe that the Terracotta discovery/replication algorithm was not working properly.

Comments

Fiona OShea 2010-12-03

Please take a look when you get a chance. It might just be that he isn’t waiting long enough (or a bug)

Fiona OShea 2010-12-03

QA Team, can you try to reproduce this issue? then assign back to DRB . thanks

Gautam Jayaprakash 2010-12-09

Kalai, can you take a look?