Feature/dynamic cluster #199
Conversation
…eblogic-kubernetes-operator into feature/dynamic-cluster
…for admin server, removed per server nodemgr_home, add dynamic cluster size verification in RestBackendImpl
…kubernetes-operator into feature/dynamic-cluster
…tartup script, no longer create machines for dynamic servers in create domain job
… with managedServerStartCount smaller than managedServerCount
… clusters. use named edit session in start-server.py when updating NodeMangerMBean
…eblogic-kubernetes-operator into feature/dynamic-cluster
Conflicts: operator/src/main/java/oracle/kubernetes/operator/http/HttpClient.java
| if (nodePort != null) { | ||
| servicePort.setNodePort(nodePort); | ||
| } | ||
| ports.add(servicePort); | ||
|
|
There was a problem hiding this comment.
Why are you exposing the node manager port here if we didn't before?
There was a problem hiding this comment.
Sorry, this was leftover code when we were trying to address the WLDF delay issue by exposing nm port that I forgot to remove.
| @@ -16,16 +16,77 @@ | |||
| import java.util.Map; | |||
|
|
|||
| /** | |||
| * Contains configuration for a WebLogid Domain | |||
| * Contains a snapsot of configuration for a WebLogic Domain | |||
| * @return true if the request to update the cluster size is successful, false if it was not successful within the | ||
| * time period, or the cluster is not a dynamic cluster | ||
| */ | ||
| public boolean updateDynamicClusterSize(final WlsClusterConfig wlsClusterConfig, |
There was a problem hiding this comment.
Who calls this? We should discuss... Other places use a pattern where this is wrapped in a Step which suspends until the asynchronous activity finishes. This allows for timeout and retry, etc.
There was a problem hiding this comment.
There are no usages. It was added some time ago when we thought the RestBackEndImpl class may need to adjust the dynamic cluster size. I will remove this method.
| set('ServerTemplate', st1) | ||
| set('ServerNamePrefix', "%MANAGED_SERVER_NAME_BASE%") | ||
| set('DynamicClusterSize', number_of_ms) | ||
| set('MaxDynamicClusterSize', 800) |
There was a problem hiding this comment.
Remove commented out set calls. Should we remove setting MaxDynamicClusterSize?
There was a problem hiding this comment.
Oh! I missed that. What's the implication of MaxDynamicClusterSize vs. DynamicClusterSize? Adding @tbarnes-us.
There was a problem hiding this comment.
In effect, D-C-S sets the number of ms that show up in the bean tree (virtual config beans of some kind), which in effect designates the servers that can be booted. For example, when DCS is 10, you will see 10 managed servers in the config tree on the console even when they aren't booted. It also can do 'double duty' of letting the wlst-scaling/external frameworks know the number of servers you'd like to be running. Max DCS, as far as I know, simply prevents DCS from being configured higher than the specified value - I personally don't think it has much utility. JMS and JTA usually require that DCS never be lowered during a scale down (instead just shutdown the servers you don't want) -- this lets these services iterate over the aforementioned 'virtual' beans to locate data that may be trapped in servers that are no longer going to be booted.
There was a problem hiding this comment.
This is being replaced with set('MaxDynamicClusterSize', number_of_ms). Also will be adding some validation logic to compare replicas and max dynamic cluster size in WlsClusterConfig class and make sure the max dynamic cluster size is honored and we won't scale above it.
| @@ -633,12 +639,23 @@ protected V1Pod computeManagedPodConfig(TuningParameters configMapHelper, Packet | |||
| V1ContainerPort containerPort = new V1ContainerPort(); | |||
| containerPort.setContainerPort(scan.getListenPort()); | |||
| containerPort.setProtocol("TCP"); | |||
| // containerPort.setName("weblogic"); // commented out as we are not exposing node manager port | |||
There was a problem hiding this comment.
Cleanup commented out section for exposing NM port
|
|
||
| boolean result = false; | ||
| // Create machine(s) | ||
| // Commented out as we are not configuring machines for servers |
There was a problem hiding this comment.
Remove since we're not creating machines implicitly anymore?
There was a problem hiding this comment.
Right. Will also be removing some method arguments and utility class related to machine creation.
…ialize it to 800. Remove machine creation related code
…eblogic-kubernetes-operator into feature/dynamic-cluster
Uh oh!
There was an error while loading. Please reload this page.