Separate the exception for transform and indexing for consuming records #7926
Conversation
| canTakeMore = _realtimeSegment.index(transformedRow, msgMetadata); | ||
| indexedMessageCount++; | ||
| realtimeRowsConsumedMeter = | ||
| _serverMetrics.addMeteredTableValue(_metricKeyName, ServerMeter.REALTIME_ROWS_CONSUMED, 1, |
There was a problem hiding this comment.
One consumed row can transform into multiple transformed rows, so we should bump this metric after we have consumed a row (and before we transform it) -- whether there are exceptions or not.
There was a problem hiding this comment.
This keeps the existing behavior where each transformed row bumps this metric ones. I feel we should keep the existing behavior
There was a problem hiding this comment.
Ah I see now. In the original code, we bump metrics independently. if multiple transformed rows, then once for each transformed row, otherwise once for consumed row. Your new code will preserve the same behavior yes. thanks for the clarification.
Codecov Report
@@ Coverage Diff @@
## master #7926 +/- ##
============================================
- Coverage 71.15% 71.06% -0.10%
- Complexity 4111 4114 +3
============================================
Files 1593 1593
Lines 82365 82375 +10
Branches 12270 12271 +1
============================================
- Hits 58609 58536 -73
- Misses 19806 19892 +86
+ Partials 3950 3947 -3
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
Currently the exception from transforming and indexing the records are mixed together, which is hard for debugging. This PR splits them into different exception messages.