mcvsubbu commented on a change in pull request #5745:
URL: https://github.com/apache/incubator-pinot/pull/5745#discussion_r460214241
##########
File path:
pinot-broker/src/main/java/org/apache/pinot/broker/broker/helix/HelixBrokerStarter.java
##########
@@ -235,7 +234,7 @@ public void start()
FunctionRegistry.init();
_brokerRequestHandler =
new SingleConnectionBrokerRequestHandler(_brokerConf, _routingManager,
_accessControlFactory, queryQuotaManager,
- _brokerMetrics, _propertyStore);
+ _brokerMetrics, new TableCache(_propertyStore));
Review comment:
@ianvkoeppe like we spoke:
Controller API that updates time boundary is probably the best solution.
Offline push jobs must pay attention to table configs so that they can do other
things that are needed for the table (partitioning, inverted index generation,
etc.). It is possible for the push job to invoke the controller API to push the
time boundary. This will also help in use cases that push multiple segments so
that the time boundary is advanced after all segments are pushed.
In addition to this, the brokers must also realize on their own that the
time boundary should be moved forward for these tables. This is because a
broker can get restarted after the push, and go into old behavior. The broker
can do this on the basis of the table config when it recognzes it has to serve
a table. We alreeady read the table config to get the quota information in the
broker.
We should not be changing the query. Instead, update the time boundary.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@pinot.apache.org
For additional commands, e-mail: commits-h...@pinot.apache.org
