Github user aledsage commented on a diff in the pull request:

    https://github.com/apache/incubator-brooklyn/pull/782#discussion_r36026516
  
    --- Diff: 
software/nosql/src/main/java/brooklyn/entity/nosql/redis/RedisClusterImpl.java 
---
    @@ -20,49 +20,73 @@
     
     import java.util.Collection;
     
    +import brooklyn.enricher.Enrichers;
     import brooklyn.entity.Entity;
     import brooklyn.entity.basic.AbstractEntity;
     import brooklyn.entity.basic.Entities;
     import brooklyn.entity.group.DynamicCluster;
     import brooklyn.entity.proxying.EntitySpec;
     import brooklyn.entity.trait.Startable;
    +import brooklyn.event.AttributeSensor;
    +import brooklyn.event.basic.Sensors;
     import brooklyn.location.Location;
     
    +import com.google.common.collect.ImmutableList;
    +import com.google.common.collect.ImmutableMap;
    +
     public class RedisClusterImpl extends AbstractEntity implements 
RedisCluster {
     
    -    protected RedisStore master;
    -    protected DynamicCluster slaves;
    +    private AttributeSensor<RedisStore> MASTER = 
Sensors.newSensor(RedisStore.class, "redis.master");
    +    private AttributeSensor<DynamicCluster> SLAVES = 
Sensors.newSensor(DynamicCluster.class, "redis.slaves");
     
         public RedisClusterImpl() {
         }
     
         @Override
         public RedisStore getMaster() {
    -        return master;
    +        return getAttribute(MASTER);
         }
         
         @Override
         public DynamicCluster getSlaves() {
    -        return slaves;
    +        return getAttribute(SLAVES);
         }
         
         @Override
         public void start(Collection<? extends Location> locations) {
    --- End diff --
    
    If you go down that road of setting starting/running, then I'd probably 
also separate it out into the two methods - delegate to something like 
`doStart()` to actually invoke start on the entities.
    
    I think you're free to instantiate the `slave` cluster before calling 
master.start. That will mean the web-console will show the master + slaves in 
the tree immediately, rather than waiting for the master to start before the 
cluster appears.
    
    You could also move the instantiation of the master + slave entities into 
init. Then you wouldn't need to guard for null either (and don't need to call 
`Entities.manage()` when inside init - on completion of init, any child 
entities will immediately get managed).


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---

Reply via email to