milleruntime commented on a change in pull request #2293:
URL: https://github.com/apache/accumulo/pull/2293#discussion_r733636869
##########
File path:
server/gc/src/main/java/org/apache/accumulo/gc/GarbageCollectionAlgorithm.java
##########
@@ -217,11 +221,43 @@ private void confirmDeletes(GarbageCollectionEnvironment
gce,
throw new RuntimeException(
"Scanner over metadata table returned unexpected column : " +
entry.getKey());
}
+ Set<String> tableIdsAfter = gce.getTableIDs();
+ ensureAllTablesChecked(Collections.unmodifiableSet(tableIdsBefore),
+ Collections.unmodifiableSet(tableIdsSeen),
Collections.unmodifiableSet(tableIdsAfter));
confirmDeletesFromReplication(gce.getReplicationNeededIterator(),
candidateMap.entrySet().iterator());
}
+ @VisibleForTesting
+ /**
+ *
+ */
+ protected void ensureAllTablesChecked(Set<String> tableIdsBefore,
Set<String> tableIdsSeen,
+ Set<String> tableIdsAfter) {
+
+ // if a table was added or deleted during this run, it is acceptable to not
+ // have seen those tables ids when scanning the metadata table. So get the
intersection
+ Set<String> tableIdsMustHaveSeen = new HashSet<>(tableIdsBefore);
+ tableIdsMustHaveSeen.retainAll(tableIdsAfter);
+
+ if (tableIdsMustHaveSeen.isEmpty() && !tableIdsSeen.isEmpty()) {
+ // we saw no table ids in ZK but did in the metadata table. This is
unexpected.
+ throw new RuntimeException(
+ "Saw no table ids in ZK but did see table ids in metadata table: " +
tableIdsSeen);
+ }
+
+ // From that intersection, remove all the table ids that were seen.
+ tableIdsMustHaveSeen.removeAll(tableIdsSeen);
Review comment:
Maybe in 2.x we need a configurable mode for the GC to handle both use
cases. One mode could be what @ivakegg desires, where the GC is more risk
adverse on a cluster that doesn't delete tables very often but is very active
with splitting and removing data. The second mode could be more aggressive
garbage collection for a cluster that is constantly adding and deleting tables.
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
