A powerful and scalable analytics engine built on top of Hadoop and HDFS. The Compute Engine is responsible for the aggregation of the status results and the computation of availability and reliability of composite services using customer defined algorithms.
Noticed that when by human declarations changes are made on the topology those changes take time to propagate to consumer logs.
For e.g. on the same date a service_flavor for a specific host might be removed from the topology but consumer logs still contain info checks about that flavor.
In the computation engine (specifically in MyUDF.AddTopology.class) lies a check looking for such inconsistencies. Resolves to throwing an exception which breaks the computation
in f4c5635 the exception is suppressed (temporary workaround)
the configuration file contains by default production-specific values and must be manually reconfigured each time is actually deployed anywhere else (devel). Should discuss how can we handle the deployment and config