mayankshriv commented on a change in pull request #3568: Add guava cache to
cache table schema in pinot broker
URL: https://github.com/apache/incubator-pinot/pull/3568#discussion_r237705963
##########
File path:
pinot-broker/src/main/java/com/linkedin/pinot/broker/requesthandler/BaseBrokerRequestHandler.java
##########
@@ -311,6 +329,68 @@ private void validateRequest(BrokerRequest brokerRequest)
{
"Value for 'LIMIT' (" + limit + ") exceeds maximum allowed value
of " + _queryResponseLimit);
}
}
+
+ // Checks whether the query contains non-existence columns.
+ // Table name has already been verified before hitting this line.
+ String tableName = brokerRequest.getQuerySource().getTableName();
+ Schema schema = _tableSchemaCache.getIfTableSchemaPresent(tableName);
+ if (schema != null) {
+ Set<String> allColumns = getAllColumnsFromBrokerRequest(brokerRequest);
+ Set<String> copied = new HashSet<>(allColumns);
+ copied.removeAll(schema.getColumnNames());
+ if (!copied.isEmpty()) {
+ _brokerMetrics.addMeteredTableValue(tableName,
BrokerMeter.QUERY_NON_EXISTENCE_COLUMNS, 1L);
+ throw new RuntimeException("Found non-existence columns from the
query: " + copied.toString());
+ }
+ } else {
+ // If the cache doesn't have the schema, loads the schema to the cache
asynchronously.
+ _tableSchemaCache.refreshTableSchema(tableName);
+ }
+ }
+
+ /**
+ * Helper to get all the columns from broker request.
+ * Returns the set of all the columns.
+ */
+ private Set<String> getAllColumnsFromBrokerRequest(BrokerRequest
brokerRequest) {
+ Set<String> allColumns = new HashSet<>();
+ // Filter
+ FilterQueryTree filterQueryTree =
RequestUtils.generateFilterQueryTree(brokerRequest);
Review comment:
The query tree is already being generated at few places in the read path. We
should think about solving it in a way that the query is parsed and query tree
generated exactly once per query.
----------------------------------------------------------------
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
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
