[CASSANDRA-17740] BulkLoader tool initializes schema unnecessarily via streaming

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 4.1-beta1, 5.0-alpha1, 5.0
• Component/s: Tool/bulk load
• Bug Category: Degradation - Resource Management
• Severity: Normal
• Complexity: Normal
• Discovered By: Adhoc Test
• Platform: All
• Since Version: 4.1

Source Control Link

https://github.com/apache/cassandra/commit/5e39c54678b0d70145b5c699593eb9ce7fa7706a

Test and Documentation Plan

Make sure existing tests aren't disturbed.

• Assignee: Caleb Rackliffe
• Reporter: Caleb Rackliffe

Authors

Caleb Rackliffe

Reviewers

Jon Meredith

• Created: 06/Jul/22 16:26
• Updated: 12/Sep/23 13:03
• Resolved: 05/Aug/22 20:51

Description

Changes to streaming setup code for CASSANDRA-17663 mean that BulkLoader initializes the schema/system keyspace, which is not what we want in a tool. Initialization is due to a call to SystemKeyspace.getPreferredIP from the BulkLoader when it starts to transmit the SSTables from the bulk loader to the Cassandra instance.

getPreferredIP:1063, SystemKeyspace (org.apache.cassandra.db)
sendMessage:213, StreamingMultiplexedChannel (org.apache.cassandra.streaming.async)
sendControlMessage:191, StreamingMultiplexedChannel (org.apache.cassandra.streaming.async)
sendControlMessage:1033, StreamSession (org.apache.cassandra.streaming)
startStreamingFiles:1257, StreamSession (org.apache.cassandra.streaming)
prepareSynAck:802, StreamSession (org.apache.cassandra.streaming)
messageReceived:622, StreamSession (org.apache.cassandra.streaming)
run:76, StreamDeserializingTask (org.apache.cassandra.streaming)
run:30, FastThreadLocalRunnable (io.netty.util.concurrent)
run:748, Thread (java.lang)

The existing BulkLoaderTest fails to detect this as it doesn't actually connect to anything so does not initialize streaming.

Affects 4.1 and trunk, and may affect 4.0, although the 4.0 patch for CASSANDRA-17663 is different than 4.1+, and may require different mitigation.

Issue Links

• is caused by: CASSANDRA-17663 Ensure FileStreamTask cannot compromise shared channel proxy for system table when interrupted
• relates to: CASSANDRA-18370 BulkLoader tool initializes schema unnecessarily via streaming - 4.0