keith-turner commented on code in PR #4524:
URL: https://github.com/apache/accumulo/pull/4524#discussion_r1672854693


##########
core/src/main/java/org/apache/accumulo/core/fate/user/UserFateStore.java:
##########
@@ -151,6 +157,95 @@ protected void create(FateId fateId, FateKey fateKey) {
         + " and fateKey " + fateKey + " after " + maxAttempts + " attempts");
   }
 
+  @Override
+  public Optional<FateTxStore<T>> tryReserve(FateId fateId) {
+    // Create a unique FateReservation for this reservation attempt
+    FateReservation reservation = FateReservation.from(lockID, 
UUID.randomUUID());
+
+    FateMutator.Status status = 
newMutator(fateId).putReservedTx(reservation).tryMutate();
+    if (status.equals(FateMutator.Status.ACCEPTED)) {
+      return Optional.of(new FateTxStoreImpl(fateId, reservation));
+    } else if (status.equals(FateMutator.Status.UNKNOWN)) {
+      // If the status is UNKNOWN, this means an error occurred after the 
mutation was
+      // sent to the TabletServer, and it is unknown if the mutation was 
written. We
+      // need to check if the mutation was written and if it was written by 
this
+      // attempt at reservation. If it was written by this reservation attempt,
+      // we can return the FateTxStore since it was successfully reserved in 
this
+      // attempt, otherwise we return empty (was written by another reservation
+      // attempt or was not written at all).
+      try (Scanner scanner = context.createScanner(tableName, 
Authorizations.EMPTY)) {
+        scanner.setRange(getRow(fateId));
+        
scanner.fetchColumn(TxColumnFamily.RESERVATION_COLUMN.getColumnFamily(),
+            TxColumnFamily.RESERVATION_COLUMN.getColumnQualifier());
+        FateReservation persistedRes = scanner.stream()
+            .filter(entry -> 
FateReservation.isFateReservation(entry.getValue().toString()))
+            .map(entry -> 
FateReservation.from(entry.getValue().toString())).findFirst()

Review Comment:
   > Thinking about this again, it would probably be better to just 
delete/create the column as needed instead of storing a not reserved value. 
What do you think?
   
   Either way seems like it could be correct.  Personally I would prefer 
deleting.   If going with the NOT_RESERVED value approach it would be good to 
make the validation of the column value more strict, will make a comment about 
that elsewhere.



-- 
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]

Reply via email to