[CASSANDRA-18026] Operability improvements around internode_max_message_size[_in_bytes] in 4.0+

• Type: Improvement
• Status: Open
• Priority: Normal
• Resolution: Unresolved
• Fix Version/s: 4.0.x, 4.1.x, 5.x
• Component/s: Documentation/NEWS.txt, Local/Config, Messaging/Internode
• Change Category: Operability
• Complexity: Normal
• Platform: All
• Assignee: Unassigned
• Reporter: Caleb Rackliffe
• Created: 09/Nov/22 21:39
• Updated: 07/Mar/23 10:54

Description

There are a few things we could do to make internode_max_message_size (formerly internode_max_message_size_in_bytes) a bit more friendly:

1.) If a read fails because of an OversizedMessageException, it would be nice to emit a client warning. Currently the client gets a ReadFailureException, surfaced as something like this:

msg="Cassandra failure during read query at consistency QUORUM (3 responses were required but only 0 replica responded, 2 failed)"
query="SELECT key, column1, value FROM ks.tbl WHERE key=? AND column1>? AND column1<=? LIMIT 256;"
stackTrace="com.datastax.driver.core.exceptions.ReadFailureException: Cassandra failure during read query at consistency QUORUM (3 responses were required but only 0 replica responded, 2 failed)

2.) We should add an entry in NEWS.txt to indicate the possible impact of internode_max_message_size_in_bytes when upgrading to 4.0+.

3.) internode_max_message_size_in_bytes / internode_max_message_size should be settable at runtime via JMX/vtable/etc.