Northflank

GPUs

Features

Resources

Changelog

Pricing

← Back to Changelog

4th July 2022

June Changelog - Read replicas, networking and volumes

Welcome to Northflank's June changelog. We've been working on a wide variety of product and developer experience enhancements across the platform. For production workloads improvements to read replicas and networking are most notable.

June's improvements and fixes:

  • Added HOST_READ connection detail to multi-replica MySQL and Postgres addons
  • Added platform handling for volume resize cooldowns on AWS clusters
  • Added support for overriding image entrypoints providing full flexibility for starting arbitrary processes within the container. This also adds full support for multiple process types in buildpack images that use Procfiles
  • Added team invite components onto entity dashboards and on-boarding flow allowing developers to jump right into their teams
  • Added support for MySQL global transaction ID (GTID) mode. GTID mode is now activated by default and improves read replication stability
  • Added ability to disable routing on default Northflank domains when custom domains are added to a port
  • Added query parameter syncing across team multiplayer, ensuring team filtering was correctly propagated
  • Added icons to notification integration event selection
  • Disabled TLS 1.0 and TLS 1.1 to improve SSL rating and security
  • Improved addon connection details with external connection details now highlighted and increased search capabilities
  • Improved domain lists with reactivity to external changes from other uses
  • Improved handling of users, group and volume permissions: Workloads are started with the correct user group if the image defines a non-root user which fixes several issues with volume permissions for buildpack-based images. Also, the metadata of files on persistent volumes will be automatically changed such that their group owner matches the group of the image’s main process ensuring that workloads can always access attached persistent volumes.
  • Improved service mesh TCP idle timeouts for internal and external endpoints allowing connections to be maintained for longer than 60 minutes
  • Improved handling of unexpected error codes from self-hosted providers, and are now displayed to the user rather than being suppressed to make it clearer when there is an issue with the provider
  • Improved performance when handling git cloning for template repositories with use of a queue
  • Improved developer experience of the service creation and type selection
  • Improved stability of metrics-server inside Kubernetes to improve platform stability
  • Improved MySQL dump imports. Importing a dump will now import to the default database if no database is specified in the dump
  • Improved account limits UI to show current usage and percentage of total limit
  • Updated default branch rule for build services to be main in cases where the default branch has not been retrieved from a repository
  • Improved global docker registry caching to better work with Kaniko and Buildkit cache
  • Improved latency and responsiveness of environment editors
  • Fixed creating a volume during service creation for teams
  • Fixed environment editor crash on null secrets array
  • Fixed issue with volume being marked as deleted before being provisioned
  • Fixed volume configuration allowing selection of volume size smaller than current volume configuration
  • Fixed issue with utf-8 encoding for secret files and config maps
  • Fixed an issue with GitLab account links where builds could in rare cases fail to be triggered if multiple users with the same account linked triggered builds simultaneously
  • Fixed Ruby on Rails template
  • Fixed cron schedule text to show a 24 hour clock instead of 12 hour

Guides

API, JS-client and CLI:

  • Added support for setting Docker entrypoint overrides during service and job creation
  • Added Set service Docker entrypoint override and Set job Docker entrypoint override endpoints to modify Docker entrypoint override after resource creation
  • Update job deployment endpoint can now be called without passing in a build service id, which will cause it to keep using the existing build service.
  • Improved performance of Update service runtime environment endpoint to make response time faster
  • Fixed an issue with Scale service endpoint where scaling to zero instances would not update correctly.
  • Improved error handling when linking an addon to a secret group to make it clearer which keys are valid for each addon
  • Fixed issue with Update job settings endpoint where runOnSourceChange setting would sometimes not update correctly

Documentation

Share this article with your network

Next product developments