This message was deleted.
# troubleshootings
Slackbot
05/04/2023, 3:54 AMThis message was deleted.
👍 2
d
Didip Kerabat
05/04/2023, 4:34 AMI have to note, this only works if you put hostname inside druid.host. Don’t put IP addresses.
f
Frank Cao
05/04/2023, 6:18 AM@Didip Kerabat This is really cool. Thanks for your work. Can I ask why we need to have two clusters to achieve 0 downtime environment? Per the Druid document, it is suggesting to use rolling updates.
d
Didip Kerabat
05/04/2023, 5:02 PM@Frank Cao This stems from the need to roll upgrade AWS EKS versions. Druid has never been able to tolerate EKS rolling upgrade (Druid 25 dealt with it much better thanks to the new Apache Curator library), but it wasn’t perfect. For some of our apps, they are P0 mission critical. We must absolutely achieve 0 downtime.
Didip Kerabat
05/04/2023, 6:48 PMOne more thing, you need to make your historical tiers looked like the following so that you can replicate between 2 clusters. This cannot be achieved using the UI. You have to submit directly using the API.
2 Views