Open kortemik opened 8 months ago
see pth_10-issue-197 in confluence @StrongestNumber9
Any reference performance numbers or environment specs? Is it 10% slower, 50%, 90%, 99%? And how much data (event count and roughly event size) just so I can test theoretical speeds of raw hdfs copy as well
at least 50? also the data was not forwarded to the second destination specified on the query as it used to.
copy is always faster than avro appends so it does not make sense to test the copy speed compared to file structured write.
internal pr request for kafka fix submitted
Describe the bug hdfs save performance regression reported since 3b211fee33aa6257f106395097cdf8c9cb14f198
Expected behavior hdfs save performance should be better or equal as it was before.
How to reproduce use hdfs save (possibly without aggregates)
Screenshots
mode is possibly the culprit.
Software version 4.16.0
Desktop (please complete the following information if relevant):
Additional context