[CASSANDRA-8512] cqlsh unusable after encountering schema mismatch

• Type: Bug
• Status: Resolved
• Priority: Low
• Resolution: Fixed
• Fix Version/s: 2.1.3
• Component/s: Legacy/Tools
• Labels: cqlsh

Environment

Cassandra 2.1.2

• Severity: Low
• Assignee: Tom Hobbs
• Reporter: Jeff Brewster

Authors

Tom Hobbs

Reviewers

Aleksey Yeschenko

• Tester: Philip Thompson
• Created: 18/Dec/14 16:26
• Updated: 16/Apr/19 09:31
• Resolved: 02/Jan/15 18:15

Description

When starting cqlsh against a node that reports a schema mismatch, no metadata is available and all queries fail.

3-node cluster
2 nodes report schema mismatch when connecting to them via cqlsh (see attached cqlsh-debug.txt)
schema mismatch is with a 4th node that was previously started, failed to bootstrap, and then removed from the cluster

nodetool status:
UN 10.218.148.82 1.22 GB 256 ? 0d6a40ed-47d9-4d95-8e3d-cf7f82d69512 rack1
UN 10.237.206.203 2.31 GB 256 ? e75cd179-e13e-462f-ba34-4c9c08a4e529 rack1
UN 10.33.182.183 2.4 GB 256 ? 5a2875d5-e736-4afd-845e-635df9bc4731 rack1

nodetool describecluster:
Cluster Information:
Name: ClusterName
Snitch: org.apache.cassandra.locator.DynamicEndpointSnitch
Partitioner: org.apache.cassandra.dht.Murmur3Partitioner
Schema versions:
383890fa-5879-32bb-b3d3-a3e3c1648ecf: [10.218.148.82, 10.237.206.203, 10.33.182.183]

Issue Links

• relates to: CASSANDRA-9689 keyspace does not show in describe list, if create query times out