[PHOENIX-1650] Possible resource leak in upsert select query

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: 4.3.0
• Assignee: Rajeshbabu Chintaguntla
• Reporter: Rajeshbabu Chintaguntla
• Created: 11/Feb/15 00:10
• Updated: 21/Nov/15 02:15
• Resolved: 11/Feb/15 00:55

Description

In the upsert select query execution path if parallelIteratorFactory then we are not closing the iterator which results in resource leak.

                    ResultIterator iterator = queryPlan.iterator();
                    if (parallelIteratorFactory == null) {
                        return upsertSelect(statement, tableRef, projector, iterator, columnIndexes, pkSlotIndexes);
                    }
                    parallelIteratorFactory.setRowProjector(projector);
                    parallelIteratorFactory.setColumnIndexes(columnIndexes);
                    parallelIteratorFactory.setPkSlotIndexes(pkSlotIndexes);
                    Tuple tuple;
                    long totalRowCount = 0;
                    while ((tuple=iterator.next()) != null) {// Runs query
                        Cell kv = tuple.getValue(0);
                        totalRowCount += PLong.INSTANCE.getCodec().decodeLong(kv.getValueArray(), kv.getValueOffset(), SortOrder.getDefault());
                    }
                    // Return total number of rows that have been updated. In the case of auto commit being off
                    // the mutations will all be in the mutation state of the current connection.
                    return new MutationState(maxSize, statement.getConnection(), totalRowCount);