[HDFS-17359] EC: recheck failed streamers should only after flushing all packets.

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 3.4.0, 3.3.9, 3.4.1, 3.5.0
• Component/s: ec
• Labels: pull-request-available
• Assignee: farmmamba
• Reporter: farmmamba
• Created: 26/Jan/24 12:27
• Updated: 11/Feb/24 01:10
• Resolved: 01/Feb/24 15:01

Description

In method DFSStripedOutputStream#checkStreamerFailures, we have below codes:

    Set<StripedDataStreamer> newFailed = checkStreamers();
    if (newFailed.size() == 0) {
      return;
    }    if (isNeedFlushAllPackets) {
      // for healthy streamers, wait till all of them have fetched the new block
      // and flushed out all the enqueued packets.
      flushAllInternals();
    }
    // recheck failed streamers again after the flush
    newFailed = checkStreamers(); 

We should better move the re-check logic into if condition to reduce useless invocation.

Issue Links

• links to: GitHub Pull Request #6503