Github user keith-turner commented on a diff in the pull request:
https://github.com/apache/accumulo/pull/224#discussion_r106906812
--- Diff:
core/src/main/java/org/apache/accumulo/core/client/summary/Summary.java ---
@@ -78,27 +80,38 @@ public long getLarge() {
}
/**
+ * @return The number of files that were deleted after the summary
retrieval operations started. This is a rare race condition where a compaction
causes a
+ * file to be deleted while retrieving summaries. When this
happens, the file that replaced the deleted file can not be used because it may
contain
+ * duplication summary information for other files. Avoiding
this race condition would be expensive, so reporting it was chosen. If this
condition
+ * must be avoided, then compactions must be stopped.
Compactions could be stopped on a cloned table to avoid this.
+ */
+ public long getDeleted() {
+ return deleted;
--- End diff --
It was not onerous to compute. Special action needs to be taken on file
not found anyway (do not want to retry). In addition to ignoring the file,
just need to report the info back. Reporting back was not too bad as there
were already other file stats.
I could not figure out to write a test for it. I tested it manually.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
