slfan1989 commented on code in PR #13831:
URL: https://github.com/apache/iceberg/pull/13831#discussion_r2287550358
##########
flink/v2.0/flink/src/main/java/org/apache/iceberg/flink/maintenance/operator/DeleteFilesProcessor.java:
##########
@@ -76,6 +79,10 @@ public void open() throws Exception {
taskMetricGroup.counter(TableMaintenanceMetrics.DELETE_FILE_FAILED_COUNTER);
this.succeededCounter =
taskMetricGroup.counter(TableMaintenanceMetrics.DELETE_FILE_SUCCEEDED_COUNTER);
+ this.deleteFileTimeMsHistogram =
+ taskMetricGroup.histogram(
+ TableMaintenanceMetrics.DELETE_FILE_TIME_MS_HISTOGRAM,
+ new DescriptiveStatisticsHistogram(1000));
Review Comment:
@pvary @Guosmilesmile We are currently investigating the integration of the
ODS layer with Iceberg, though we do not yet have a concrete business scenario.
Nevertheless, I believe adding a metric for the duration of delete operations
is valuable: it introduces very low overhead but can significantly enhance
system observability, helping us identify potential bottlenecks and analyze
performance trends. For example, if delete times increase abnormally, we can
further investigate whether the issue stems from the underlying object
storage/HDFS performance or slower HMS responses.
In this sense, latency histograms provide a more sensitive signal than
simply tracking success/failure counts. Even if deletes succeed, a spike in
latency can quickly reveal hidden storage or service bottlenecks, which would
otherwise be difficult to detect.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org
For additional commands, e-mail: issues-h...@iceberg.apache.org
