← Back to Changelog

28th August 2023

New Infrastructure Alerts & Improved Addon Performance

Today’s changelog features exciting updates to infrastructure alerts, addons, and DX on Northflank.

Infrastructure Alerts

Infrastructure alerts are now available for resource usage events. You can add new alert types from the notification settings here to be alerted via webhook, Discord, Teams, or Slack. You can also configure infrastructure alert settings here. This makes it easier to monitor your workloads and manage resource requirements, allowing you to have peace of mind that your production workloads are running as expected.

New alerts include:

  • Events for CPU and Memory usage
  • 90% Usage Spike events when a resource hits the 90% usage threshold for a short amount of time.
  • 90% Usage Sustained events when a resource has been over the threshold for five minutes or more.

Addons

Addons received several improvements to performance and configurability. These improvements mean your Addons will be more responsive to user initiated actions and faster to create and scale, leading to a much improved user experience.

These include:

  • Faster TLS certificate generation
  • Improved reaction time to actions like backup triggers
  • Faster MongoDB replica creation through a new disk cloning mechanism
  • Support for the following stateful workload versions:
    • Redis: 7.0.11, 6.2.12
    • MySQL: 8.0.33
    • MinIO: 2023.6.9
    • MongoDB: 6.0.7, 5.0.18
    • RabbitMQ: 3.10.24, 3.11.18, 3.12.0
    • PostgreSQL: 11.20.0, 12.15.0, 13.11.0

Bug fixes include:

  • Fix for issue where Postgres native backup restores were not correctly owned by the provided admin user.
  • Fix for issue where Postgres addon could not be properly created when using a custom database name with uppercase letters.

Additionally, you can now set an eviction policy on your Redis deployment. This enables fine-grained control over your Redis deployment's behaviour as it reaches its memory limit. Under the hood, you are configuring the Redis maxmemory-policy directive. Learn more about managing key eviction in Redis here.

Lastly, you can now reset your addon, completely wiping the addon of all data and returning it to its initial state. This process does not change your existing addon settings.

Services

You can now configure the horizontal autoscaling of your services by Requests Per Second, for an extra degree of control.

Bug fixes include:

  • Fix for service deletion volume check when deleting a service via the dashboard header.
  • Fix for build duration not displaying correctly.

Jobs

The cron schedule is now viewable on the cron job settings page on initial page load.

Usability

You no longer have to be an administrator for a GitHub organisation to connect with Northflank. After the organisation admin has authorised the GitHub App installation, users can return to the Northflank Git settings to finalise the link.

Bug fixes include:

  • Fix to address load failure for code editor view in certain countries (due to blocked CDN).
  • Fix to address issue with GitLab linking, where users who have linked their GitLab account to multiple teams could have their link invalidated.
  • Fix to address certain translation browser extensions causing the site to crash.
  • Minor UI fixes to the project selection modal and health check status icon.

API & CLI

API signature updates:

  • The list and get endpoints for Addons, Services, Jobs, and Secrets now return the field tags, an array of the resource tags associated with the resource.
  • The list repositories endpoint now returns the repository id field as a string for type consistency across all version control providers. Previously, GitHub and GitLab repos returned an integer while Bitbucket repos returned a string.
  • Get logs endpoints now have timestamp-based pagination, and the maximum number of log lines returned has been lifted to 1000. Additionally, when used with the Northflank CLI, get logs endpoints alert the user when no tall log lines can be returned due to hitting the maximum number.
  • Get metrics endpoints now have improved error messages for invalid api calls.

Share this article with your network