This variable ensures that Kafka resets are not allowed by default, so as to provide safeguard against accidental misconfiguration causing a cluster split (see Switching Kafka for more information on Kafka resets).

With this variable unset by default, a change in the Kafka cluster ID or the HUMIO_KAFKA_TOPIC_PREFIX will cause nodes to crash until these two values match what is in the global snapshot. This prevents the cluster from acting on the change to Kafka until an administrator assents to having a Kafka reset performed.

To intentionally perform a Kafka reset, set ALLOW_KAFKA_RESET_UNTIL_TIMESTAMP_MS to an epoch timestamp in near future (for instance now + one hour), which will make sure that the setting is automatically disabled again once the reset is complete. You can set the timestamp to a maximum of six hours into the future.

This variable only disables the Kafka check that stops the node. It does not make LogScale treat an existing global snapshot as a disaster recovery snapshot. To start a cluster from the data of another cluster, set the matching *_RECOVER_FROM_BUCKET variable, such as AZURE_RECOVER_FROM_BUCKET, and use a new storage bucket.