keith-turner closed pull request #566: Improved comments for #559 changes
URL: https://github.com/apache/accumulo/pull/566
This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below for
the sake of provenance:
As this is a foreign pull request (from a fork), the diff is supplied
below (as it won't show otherwise due to GitHub magic):
diff --git
a/server/tserver/src/main/java/org/apache/accumulo/tserver/tablet/Tablet.java
b/server/tserver/src/main/java/org/apache/accumulo/tserver/tablet/Tablet.java
index 4dc19f293d..9554b93a62 100644
---
a/server/tserver/src/main/java/org/apache/accumulo/tserver/tablet/Tablet.java
+++
b/server/tserver/src/main/java/org/apache/accumulo/tserver/tablet/Tablet.java
@@ -2471,6 +2471,22 @@ public void importMapFiles(long tid,
Map<FileRef,MapFileInfo> fileMap, boolean s
private volatile Set<DfsLogger> referencedLogs = Collections.emptySet();
private synchronized void rebuildReferencedLogs() {
+ /*
+ * Each tablet has the following sets of WALogs. While a WALog exists in
one set, garbage
+ * collection must be avoided.
+ *
+ * 1. WALogs for the active in memory map
+ *
+ * 2. WAlogs for the minor compacting in memory map
+ *
+ * 3. WAlogs for a newly minor compacted file that is being added to the
metadata table.
+ *
+ * Set 1 is currentLogs. Set 2 is otherLogs. Set 3 only exist in
referenced logs as a side
+ * effect of not calling this method in beginClearingUnusedLogs() when
otherLogs is cleared.
+ *
+ * Ensuring referencedLogs accurately tracks these sets ensures in use
walogs are not GCed.
+ */
+
Builder<DfsLogger> builder = ImmutableSet.builder();
builder.addAll(currentLogs);
builder.addAll(otherLogs);
@@ -2505,9 +2521,10 @@ public void removeInUseLogs(Set<DfsLogger> candidates) {
}
otherLogs = Collections.emptySet();
- // Do NOT call rebuildReferenedLogs() here as that could cause walogs to
be GCed before the
- // minc rfile is written to metadata table (see #539). The clearing of
otherLogs is reflected
- // in refererncedLogs when finishClearingUnusedLogs() calls
rebuildReferenedLogs().
+ // Intentionally NOT calling rebuildReferenedLogs() here as that could
cause GC of in use
+ // walogs(see #539). The clearing of otherLogs is reflected in
refererncedLogs when
+ // finishClearingUnusedLogs() calls rebuildReferenedLogs(). See the
comments in
+ // rebuildReferenedLogs() for more info.
if (unusedLogs.size() > 0)
removingLogs = true;
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
