[ARTEMIS-2660] Kubernetes Master/Slave Rolling update

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Information Provided
• Affects Version/s: 2.11.0
• Component/s: Broker
• Assignee: Unassigned
• Reporter: Yasin Koyuncu
• Created: 16/Mar/20 15:24
• Updated: 16/Mar/20 15:30
• Resolved: 16/Mar/20 15:30

Description

Hi,

 

We've a setup in kubernetes cluster having master/slave with two distinct statefulsets for each one of them. This setup is working alright, however during rolling updates since the current pods are being terminated and replaced by the new deployment, the clients get disconnected for a brief moment. This leads ultimately to failure of some of the messages being send to the broker.

 

My question is, has anyone managed to make it work for a master/slave setup for rolling update in kubernetes?

 

Thx,

Yasin