Table of Contents

Cassandra cluster node in unexpected state

Affected versions

Any versions using a Cassandra Cluster setup.

Cause

When there is a time difference between different Cassandra nodes, this causes errors of the type "Bad timestamp generated". This can cause an unexpected state to be reported for some of the nodes.

Fix

Set up NTP on all servers and restart Cassandra.

Issue description

In a Cassandra cluster, some nodes report an unexpected state ("?N") of peer nodes. For example:

Example of unexpected state Cassandra node

The Cassandra debug and system logs contain warnings similar to the following example:

WARN   [Messaging-EventLoop-3-7] 2023-01-06 17:53:02,739 NoSpamLogger.java:95 - Bad timestamp 1673020344443 generated, overriding with currentTimeMillis = 1673023982737
Propose changes