Aled Sage created BROOKLYN-285:
----------------------------------
Summary: Dangling reference in Propagator.producer: can cause
subscription to self, and infinite subscription loop
Key: BROOKLYN-285
URL: https://issues.apache.org/jira/browse/BROOKLYN-285
Project: Brooklyn
Issue Type: Bug
Affects Versions: 0.9.0
Reporter: Aled Sage
When the "producer" entity of a Propagator enricher is unmanaged, it can cause
rebind to resolve the dangling reference value as {{null}}. In 0.9.0, that is
treated as the producer being "self". This can cause an infinite loop, where
the Propagator subscribes to its own sensor, and every time its sensor changes
it resets the sensor value.
To reproduce, deploy this app:
{noformat}
location: localhost
services:
- type: org.apache.brooklyn.entity.stock.BasicApplication
brooklyn.children:
- type: org.apache.brooklyn.entity.machine.MachineEntity
{noformat}
And programmatically add a Propagator enricher to the top-level app, such as:
{noformat}
appId = "u4VDoIOa";
childId = "LPYT4g27";
sensor = org.apache.brooklyn.core.sensor.Sensors.newSensor(String.class,
"host.name");
app = mgmt.getEntityManager().getEntity(appId);
child = mgmt.getEntityManager().getEntity(childId);
app.enrichers().add(org.apache.brooklyn.api.sensor.EnricherSpec.create(org.apache.brooklyn.enricher.stock.Propagator.class)
.configure("enricher.producer", child)
.configure("enricher.propagating.inclusions",
com.google.common.collect.ImmutableList.of(sensor)));
{noformat}
Then unmanange the child entity (e.g. via the advanced tab in the web-console),
and restart Brooklyn.
The CPU goes through the roof, and the debug log shows log messages like:
{noformat}
2016-05-31 09:02:22,599 DEBUG o.a.b.c.m.i.LocalSubscriptionManager
[brooklyn-execmanager-gfOvVoKq-3]: 388000 events for subscriber
Subscription[qF9YFEBA;BasicApplicationImpl{id=u4VDoIOa}@u4VDoIOa:host.name]
{noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
