Faster Image Pulls, Deployments and More

We are excited to announce faster Docker image pulls and deployments. As Northflank scales deployments, we continue to aim for sub-second image staging times. As nodes are created, updated, rolled or migrated to different zones the pod churn can be high as thousands of pods attempt to redeploy concurrently. This leads to a traffic jam as pods fight to pull images before one another. Northflank can regularly replace and resize Kubernetes node pools for peak performance & reliability with the guarantee that workloads are spun up in seconds rather than minutes.

Other features & fixes

• Node image pull queues were modified to allow multiple images to pull their layers in parallel
• Node image pull rate limits were doubled to allow for more efficient fetching when multiple are downloaded at once
• Added help popovers to resource types on new creation forms (service, job, addon, pipeline, service from template and secret group)
• Added help popovers across secret groups
• Improved validation of external image paths when creating jobs
• Updated documentation JSON examples for API & CLI
• Fixed an issue where the pipeline UI would crash after some services are deleted
• Fixed issues with creating and restoring MySQL & PostgreSQL backups
• Fixed an issue where Redis CLI forwarding wouldn’t work as expected in some cases
• Fixed an issue where instances in terminating status would display an incorrect status icon for a couple of seconds