New env vars:
- METADATA_CHANGE_EVENT_NAME: The name of the metadata change event topic.
- METADATA_AUDIT_EVENT_NAME: The name of the metadata audit event topic.
- FAILED_METADATA_CHANGE_EVENT_NAME: The name of the failed metadata change event topic.
This will need to be consistent throughout your ecosystem.
CLOSES: #1840
* Bump up kafkaAvroSerde to support security config for Confluent Schema Registry
* Support certs in secrets
* Extra Spring config (e.g. security)
* Optional values
* Clarify log warnings
* Update faq.md
Co-authored-by: Lars Nielsen <Lars.Nielsen@kindredgroup.com>
Co-authored-by: Mars Lan <mars.th.lan@gmail.com>
The coupling was between the static path extractor API. This broken by making a new `UrnPathExtractor` interface, and adding an overload of `EbeanLocalDAO`'s constructor to accept one (no breaking constructor change). The old constructors default to an `EmptyPathExtractor`, which does nothing (which is a breaking behavioral change, see below).
BREAKING: `DatasetUrnPathExtractor` was deleted. No one should've been depending on this directly. However, downstreams that were relying on it being there at runtime (dataset GMS) need to copy `DatasetUrnPathExtractor` and create their `EbeanLocalDAO` with one. Note that this is a little dangerous becasue it is a runtime behavioral change only. Potential impact is that SCSI suddenly stops working as intended.
SYNC=metadata-models_101.0.0
- Add build files for new module.
- Correctly edit TestUtils since it isn't synced.
- Reference new test utils.
- Delete duplicate pdl files.
SYNC=metadata-models_100.0.1
Requested by a few people in OS. See https://github.com/linkedin/datahub/issues/1840.
Companies need full customization over the topic name. This new class should be easily customizable by using a spring factory.
TODO to finish the implmentation for v5. For right now v5 is LI only and unfinished. Getting this in for v4 so it is useful to other companies now.
TODO AFTER OPEN SOURCE PUSH - make configurable via spring
TODO AFTER SUBMIT - see where else we can use this (jobs, where else?)
* fix (docker): Fix install of Chrome in frontend Dockerimage
Retry installing Chrome after dependencies have been installed
* fix (docker): Install Chrome with apt-get
Install Chrome and dependencies at the same time, using apt-get
