[FLINK-24465] Wrong javadoc and documentation for buffer timeout

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.14.0, 1.13.2
• Fix Version/s: 1.13.6, 1.14.3, 1.15.0
• Component/s: Documentation, Runtime / Configuration, Runtime / Network
• Labels: pull-request-available

Release Note

Configuring buffer timeout to be equal to `-1` to flush only when buffers were full was broken in FLINK-18832. In Stream execution mode it was overwritten with 100ms. Fortunately it should not have had a visible impact as `100ms` should still be enough time to fill in buffers before sending them downstream.

• Assignee: Dawid Wysakowicz
• Reporter: Dawid Wysakowicz
• Created: 07/Oct/21 07:09
• Updated: 15/Dec/21 01:44
• Resolved: 18/Oct/21 14:52

Description

The javadoc for setBufferTimeout and similarly the documentation for execution.buffer-timeout claims:

    /**
     * Sets the maximum time frequency (milliseconds) for the flushing of the output buffers. By
     * default the output buffers flush frequently to provide low latency and to aid smooth
     * developer experience. Setting the parameter can result in three logical modes:
     *
     * <ul>
     *   <li>A positive integer triggers flushing periodically by that integer
     *   <li>0 triggers flushing after every record thus minimizing latency
     *   <li>-1 triggers flushing only when the output buffer is full thus maximizing throughput
     * </ul>
     *
     * @param timeoutMillis The maximum time between two output flushes.
     */

which is not true.

The -1 value translates to 100ms. See org.apache.flink.streaming.api.graph.StreamingJobGraphGenerator#checkAndResetBufferTimeout

Issue Links

• links to: GitHub Pull Request #17471