User Defined Function Execution
User Defined Functions (UDFs) which python can use to extend fields needs to be populated in a rule were earlier executed in each ingest. The python interpreters which are spawned to execute these UDF’s were mostly idle but could handle only one request at a time. This increased the overall processing time when using these functions. Even with enough computing power and RAM available, the rate at which HealthBot was able to consume the data was low.
These UDF executions have been moved out of TAND and into a UDF farm which can process multiple requests at a time. This has resulted in a 4.5 to 5 times improvement in the rate at which HealthBot can process rules that have UDFs.
Invalid/Missing Kubernetes Configuration During Install
An error appears during multi-node installation: Invalid/Missing kubernetes configuration.
Due to changes in back end infrastructure the HealthBot CLI
no longer has access to files outside of the path
/var/local/healthbot/. To correct this, you can
create an environment variable using the command: export HB_EXTRA_MOUNT1=/root/.kube/config, or copy the kubernetes configuration file into the
After completing one of these two, the installer will complete normally.
Holt-Winters Issue in Anomaly Detection
An issue which prevented the successful use of Holt-Winters anomaly detection and prediction in certain circumstances has been resolved.
Machine Learning Issue
In some rules where tagging was used, machine learning algorithms did not function. This issue has been resolved.
Tagged Field References
In rules that contained tagged fields, using a reference to those fields from other rules would fail. This issue has been resolved.