[BEAM-7507] StreamingDataflowWorker attempts to decode non-utf8 binary data as utf8

• Type: Improvement
• Status: Resolved
• Priority: P4
• Resolution: Fixed
• Fix Version/s: 2.14.0
• Component/s: runner-dataflow
• Assignee: Steve Niemitz
• Reporter: Steve Niemitz
• Created: 07/Jun/19 13:58
• Updated: 24/Jul/20 19:57
• Resolved: 07/Jun/19 16:19

Description

In StreamingDataflowWorker, when setting the workId for the logging MDC, currently the code attempts to decode a binary blob as utf8.  This will (generally) cause a decoding error, and the java UTF8 decoder uses a globally synchronized cache to handle generating these error results.

A simple solution would be to use the protobuf TextFormat escaping to convert the ByteString key into a string.

Issue Links

• links to: GitHub Pull Request #8789
• links to: GitHub Pull Request #8888