apache/cassandra · error · InvalidRequestException

Cannot use selection function %s on PRIMARY KEY part %s

Error message

Cannot use selection function %s on PRIMARY KEY part %s

What it means

Selection functions of certain kinds (e.g. writetime, ttl, ttl-like wrappers) cannot be applied to columns that are part of the primary key, because primary key columns have no per-row writetime/ttl cells to read. The factory refuses to build the selector.

Source

Thrown at src/java/org/apache/cassandra/cql3/selection/Selectable.java:321

        {
            this.column = column;
            this.selectable = selectable;
            this.kind = kind;
        }

        @Override
        public String toString()
        {
            return kind.name + "(" + selectable + ")";
        }

        public Selector.Factory newSelectorFactory(TableMetadata table,
                                                   AbstractType<?> expectedType,
                                                   List<ColumnMetadata> defs,
                                                   VariableSpecifications boundNames)
        {
            if (column.isPrimaryKeyColumn())
                throw new InvalidRequestException(
                        String.format("Cannot use selection function %s on PRIMARY KEY part %s",
                                      kind.name,
                                      column.name));

            Selector.Factory factory = selectable.newSelectorFactory(table, expectedType, defs, boundNames);
            boolean isMultiCell = factory.getColumnSpecification(table).type.isMultiCell();

            return WritetimeOrTTLSelector.newFactory(factory, addAndGetIndex(column, defs), kind, isMultiCell);
        }

        @Override
        public AbstractType<?> getExactTypeIfKnown(String keyspace)
        {
            AbstractType<?> type = kind.returnType;
            return column.type.isMultiCell() && !kind.aggregatesMultiCell() ? ListType.getInstance(type, false) : type;
        }

        @Override

View on GitHub (pinned to 88fd0f6a0e)

Solutions

1. Remove the selection function from the primary key column
2. Apply the function to a regular (non-key) column
3. If you need key write history, model it in a separate table with the key as a value column

Example fix

// before
SELECT writetime(pk) FROM t;
// after
SELECT writetime(value) FROM t;

Set<String> pkCols = table.primaryKeyColumns().stream().map(ColumnMetadata::getName).collect(toSet());
if (pkCols.contains(targetColumn)) throw new IllegalArgumentException("writetime/ttl not allowed on PK column");

When it happens

Trigger: SELECT writetime(pk_col) FROM t or TTL(clustering_col) where the wrapped column belongs to the PRIMARY KEY.

Common situations: Users probe write timestamps of partition/clustering keys; Cassandra never stores writetime/ttl for primary key columns, so the query is rejected at prepare time.

Understand the failure class

Background: UnsupportedOperationException and "is not supported" errors: when a library deliberately refuses a call — this error's family across 30 libraries.

Related errors

• %s constraint can not be specified on a %s key column '%s'
• PRIMARY KEY column "%s" cannot be restricted (preceding colu
• Non PRIMARY KEY columns found in where clause: %s
• Cannot drop PRIMARY KEY column %s
• Unknown column '%s' referenced in PRIMARY KEY for table '%s'

AI-assisted analysis of apache/cassandra@88fd0f6a0e (2026-09-10). Data as JSON: /api/errors/73788973a19574ac. Report an issue: GitHub.