- Add debug path for fetching Vizier logs

- Fix memory leak in Kelvin.
- Propagate NATS size errors to UI.

- Fix bug with metadata service where PebbleDB was polled too frequently.
- Data collector optimizations.

- Update cgroup paths to support CRI-O.

- Data collector fixes for OpenShift environments

- Switch to pebbleDB for metadata store as default.
- Better messaging for script execution errors.

- Display K8s events for Vizier control plane pods in admin UI.
- Add toleration on PEM daemonset to allow for deployment on tainted nodes.

- Increase leader election timeout to reduce leadership failures.

- Improve metadata update distribution scheme, by removing metadata updates
from heartbeats

- Cleanup in data collector and metadata.

- Data collector performance optimizations

- Fix metadata bug for Openshift clusters.

- Fix metadata init-container bug for older Viziers.

- Fix memory corruption bug in data collector which was causing frequent PEM crashes

- Add ip_to_service_id UDF
- Fix OpenSSL tracing bug - allow capture of multiple requests from Ruby
clients
- Fix segfault bug for HTTP2 tracing

- Fix bug with DNS tracing

- Support Redis pub/sub tracing.
- Add bytes_to_hex udf.

- Bug fixes for displaying semantic types in results

- Data collector code cleanup

- Add init container to metadata pod, to wait for etcd.
- Log cleanup.