Bug #119627 MySQL 8.0.44 K8s cluster (using the operator) comes up with Bad handshake error messages on primary cluster member

• Submitted: 5 Jan 16:50
• Modified: 5 Jan 16:52
• Reporter: Ronny Bremer
• Status: Open
• Category: MySQL Operator
• Severity: S3 (Non-critical)
• Version: 9.5.0-2.2.6
• OS: Red Hat
• CPU Architecture: x86

[5 Jan 16:50] Ronny Bremer

Description:
I deployed the MySQL operator helm chart version 2.2.6 in a K8s cluster on premise. I can successfully deploy InnoDB clusters with MySQL version 8.4.7 and 9.5.0, however, when using version 8.0.44 the cluster comes up with those logs:
2026-01-05T16:26:11.282845Z 3527 [Note] [MY-010914] [Server] Bad handshake
2026-01-05T16:26:12.669040Z 3529 [Note] [MY-010914] [Server] Bad handshake
2026-01-05T16:26:12.935231Z 3530 [Note] [MY-010914] [Server] Bad handshake
2026-01-05T16:26:12.969233Z 3531 [Note] [MY-010914] [Server] Bad handshake
2026-01-05T16:26:13.163166Z 3532 [Note] [MY-010914] [Server] Bad handshake
2026-01-05T16:26:13.298386Z 3533 [Note] [MY-010914] [Server] Bad handshake
2026-01-05T16:26:14.054287Z 3534 [Note] [MY-010914] [Server] Bad handshake
2026-01-05T16:26:14.982918Z 3535 [Note] [MY-010914] [Server] Bad handshake
2026-01-05T16:26:15.183654Z 3536 [Note] [MY-010914] [Server] Bad handshake
on the primary cluster node. I tried re-deploying multiple times, to no avail.

How to repeat:
deploy helm chart mysql-operator (version 9.5.0-2.2.6) in a K8s cluster.
deploy helm chart mysql-innodbcluster with self signed certificates and serverVersion set to 8.0.44
wait for the cluster to deploy
observe the Bad handshake messages on pod cluster-0

[5 Jan 16:52] Ronny Bremer

For reference, this is the yaml file used for the InnoDB Cluster helm chart deploy:

credentials:
  root:
    user: root
    password: "**redacted**"
    host: "%"

tls:
  useSelfSigned: true

routerInstances: 1
serverInstances: 2
serverVersion: 8.0.44

serverConfig:
  mycnf: |
    [mysqld]
    max_connections=1000
    transaction-isolation=READ-COMMITTED
    max_allowed_packet=265M
    sql_mode="ONLY_FULL_GROUP_BY,STRICT_TRANS_TABLES,ERROR_FOR_DIVISION_BY_ZERO,NO_ENGINE_SUBSTITUTION"

datadirVolumeClaimTemplate:
  resources:
    requests:
      storage: 50Gi