I've been tailing the server.log in vRO 7.2, using the Tech Preview vCenter 6.5.0.5817435 plugin, and noticed that it might be doing some kind of proactive caching of objects from what looks to be SDK connections that have recently been used. I say its proactive, because even after workflows have ended for a specific SDK connection, I still see vRO retrieving properties for objects I've just queried. Is this normal? This definitely does speed up workflows that traverse SDK connections, since some objects are cached, but the logging does go a bit nuts.
For instance, I have a workflow that takes a DRS cluster as input. It then traverses each host in the cluster and finds all VMs on each host, then eventually spits out an analysis of the vCPU + memory usage. The last execution of this workflow ended at 15:36:26 today:
2017-07-24 15:36:26.893+0000 [WorkflowExecutorPool-Thread-9] INFO {jchou@domain.net:Get Cluster Subscription Ratios:86503545-4851-45fd-a930-34d92f4729c9:token=8a828a815d74fec4015d7538803c0045} [WorkflowHandler] End of workflow 'Get Cluster Subscription Ratios' (8a828a815d74fec4015d7538803c0045), state: completed
But the server.log still shows vRO retrieving objects from that SDK connection afterwards. When I look at server.log after 15:36:26, I see thousands of object retrieval requests for the same SDK connection. Not sure how these are triggered?
2017-07-24 15:36:28.970+0000 [https-jsse-nio-0.0.0.0-8281-exec-7] INFO {} [ObjectRetrieverImpl] Object retrieving started in user@vcenter SDK of properties [] for type VirtualMachine and number of objects: 1
...
2017-07-24 15:39:31.051+0000 [https-jsse-nio-0.0.0.0-8281-exec-6] INFO {} [ObjectRetrieverImpl] Object retrieving started in user@vcenter SDK of properties [] for type ClusterComputeResource and number of objects: 1
From 15:36:28 until 15:39:31, there are 2243 requests for objects from this SDK connection: 2155 for VirtualMachine, 80 for HostSystem, and 8 for ClusterComputeResource. To me this seems excessive, but is this how it's supposed to work? I don't recall seeing this level of logging in earlier vCenter plugins in vRO 7.2.
Is someone able to comment if this is normal?