Aled Sage created BROOKLYN-202:
----------------------------------
Summary: managementPlaneId should not change on server restart
Key: BROOKLYN-202
URL: https://issues.apache.org/jira/browse/BROOKLYN-202
Project: Brooklyn
Issue Type: Bug
Affects Versions: 0.8.0
Reporter: Aled Sage
The {{managementContext.getManagementPlaneId()}} should stay the same when a
Brooklyn server is restarted. If you have two brooklyn servers running in
high-availability mode, then they should both return the same management plane
id.
However, this is not currently true: the id changes on rebind; and in
high-availability mode then each server has a different id.
Below is the javadoc from {{ManagementContext}} described the expected
behaviour.
{noformat}
/**
* UID for the Brooklyn management plane which this {@link
ManagementContext} node is a part of.
* <p>
* Each Brooklyn entity is actively managed by a unique management plane
* whose ID which should not normally change for the duration of that
entity,
* even though the nodes in that plane might, and the plane may go down and
come back up.
* In other words the value of {@link
Application#getManagementContext()#getManagementPlaneId()}
* will generally be constant (in contrast to {@link
#getManagementNodeId()}).
* <p>
* This value should not be null unless the management context is a
non-functional
* (non-deployment) instance. */
String getManagementPlaneId();
/**
* UID for this {@link ManagementContext} node (as part of a single
management plane).
* <p>
* No two instances of {@link ManagementContext} should ever have the same
node UID.
* The value of {@link
Application#getManagementContext()#getManagementNodeId()} may
* change many times (in contrast to {@link #getManagementPlaneId()}).
* <p>
* This value should not be null unless the management context is a
non-functional
* (non-deployment) instance. */
String getManagementNodeId();
{noformat}
I encountered this with 0.9.0-SNAPSHOT. I ran {{brooklyn launch --persist auto
--persistenceDir /Users/aled/temp/empty}}, then restarted. I also then ran a
second instance (thus in "HA STANDBY" mode), and got that management plane id
via the groovy console.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
