This release improves the integration with Google Cloud Logging.
Download the release on GitHub.
Changes
Section titled “Changes”Split out name option and use metadata server when unset
Section titled “Split out name option and use metadata server when unset”We split the name option of to_google_cloud_logging, no longer requiring
user to construct the ID manually.
Bug Fixes
Section titled “Bug Fixes”Delay catalog lookups until catalog is ready
Section titled “Delay catalog lookups until catalog is ready”The export, metrics, diagnostics and partitions operators returned an
empty result when used before the node had successfully loaded its persisted
data. They now wait correctly.
By @dominiklohmann in #5156.
Always shut down the pipeline manager first
Section titled “Always shut down the pipeline manager first”We fixed a very rare bug that on shutdown could mark running pipelines as stopped, completed, or failed, causing the pipelines not to restart alongside the node.
By @dominiklohmann in #5163.
Do not refuse startup when pid.lock is invalid
Section titled “Do not refuse startup when pid.lock is invalid”The node no longer refuses to start when its last shutdown happened in the brief period on startup after its PID file was created and before it was flushed.
By @dominiklohmann in #5164.
Split out name option and use metadata server when unset
Section titled “Split out name option and use metadata server when unset”The to_google_cloud_logging operator is now available in both Docker and static
builds. The operator had earlier been missing due to a configuration issue.