Improve health signal for StreamingDataflowWorker
Open
dataflow
improvement
P3
runners
- Dominant language
- Java
- Stars
- 8.7k
- Forks
- 4.7k
- Avg merge
- 1d 20h
- Merged PRs (30d)
- 196
Description
The StreamingDataflowWorker currently just uses the health of the FnAPI server for its own health monitoring. It should also take into account persistent failures to process work items successfully which may be due to a bad worker. For example, a recent Snappy initialization error was due to /tmp becoming full.
Imported from Jira [BEAM-7796](https://issues.apache.org/jira/browse/BEAM-7796). Original Jira may contain additional context.
Reported by: scwhittle.
Contributor guide
Assessment
This issue has not been assessed yet.