Skip to content

Tags

Tags give the ability to mark specific points in history as being important
This project is mirrored from https://gitlab.com/guided-explorations/aws/gitlab-runner-autoscaling-aws-asg.git. Pull mirroring failed .
Repository mirroring has been paused due to too many failed attempts. It can be resumed by a project maintainer or owner.
Last successful update .
  • v1.5.5
    1c46d180 · Release Updates ·
    ## [v1.5.5] - 2023-07-20
    
    - ADDED new allocation strategy price-capacity-optimized as per contribution by @arichtman-srt.
    - ADDED 3GITLABRunnerDockerImage - New parameter to control default docker image for docker executors.
    - UPDATED linux docker runner for certificates and pegged version of docker container.
    - UPDATED expanded references to cinst to choco install as per [Chocolatey Deprecation of Shims](https://gitlab.com/guided-explorations/aws/gitlab-runner-autoscaling-aws-asg/-/issues/73)
    - ADDED ability to use new Runner Tokens and Legacy Runner Registration tokens. For compatibility the parameter name remains the same (3GITLABRunnerRegTokenList)
      - NOTE - as of GitLab Runner 16.2.0, the new token can be used with the legacy parameter and it will simply ignore the registration parameters that are now a fixed part of the runner authentication token generation. This is documented here: [Legacy-compatible registration processing](https://docs.gitlab.com/runner/register/#legacy-compatible-registration-processing) This means that as long as you are using runner 16.2.0 or later, you can keep using automation forked from an earlier version of this code as long as your automation does not depend on dynamic tagging or creating other parameters dynamically that are a fixed part of the new token creation process. This ability will be retired with the legacy tokens are retired (slated for 16.6.0 as of this writing)
      - NOTE: One key benefit to leveraging the new tokens is that as the ASG Scaling runner scales up and down, you no longer end up with many runner tokens being added and removed in the UI. All ASG instances now appear under one GitLab runner in the GitLab UI.
      - [READ MORE about this change](README.md#new-and-legacy-runner-tokens)
    - NEW KNOWN ISSUE: [Amazon Linux 2023 Does Not Work](https://gitlab.com/guided-explorations/aws/gitlab-runner-autoscaling-aws-asg/-/issues/74)
    - NEW KNOWN ISSUE: [Windows 2022 Does Not Work](https://gitlab.com/guided-explorations/aws/gitlab-runner-autoscaling-aws-asg/-/issues/75)
  • v1.5.0-beta1
  • v1.4.9-alpha14
    See CHANGELOG.md for full list of Changes.
  • experimental
  • v1.4.8-alpha13
  • latest
    9d1e8d23 · correction ·
    latest
  • v1.4.7-alpha12
    v1.4.7-alpha12:
    - As per AWS Spot team recommendations: 5SPOTSpotAllocationStrategy now defaults to 'capacity-optimized-prioritized' 
    - Replaced single ARM64 Easy button with two that mimic the Linux ones where the user can provide the number of instances to obtain Warm HA, Hot HA or scaling.
    - Forced 5ASGSelfMonitorTerminationInterval to lowest value (currently 1) for any easy button containing spot instances rather than rely on default value in the template
  • v1.4.6-alpha11
    - Surfaced parameters for better control of large scale updates:   4ASGUpdateMinInstancesInService and 4ASGUpdateMaxBatchSize
    - Upped maximum initially deployable instances to 20 (update your CF stack to push this higher after deployment)
    -  In [README.md](./README.md), added documentation section **GitLab Runners on AWS Spot Best Practices**
    -  In [README.md](./README.md), added a link to video: [Provisioning 100 GitLab Spot Runners on AWS in Less Than 10 Minutes Using Less Than 10 Clicks + Updating 100 Spot Runners in 10 Minutes](https://youtu.be/EW4RJv5zW4U)
  • v1.4.5-alpha10
    - Simplification of number of easy buttons provided.  Consolidated all manual scaling options and scheduling on per-platform and per-spot basis. Users pick 1, 2, or more instances to control Warm HA, Hot HA or Manual Scaled Fleet. Instance count can be tuned via ASG parameter edits after deployment.
    - Add preflight end-to-end connection tests for endpoints needed for successful installation and configuration of the runner. Fail immediately if there is a possible network problem between the runner network context and the GitLab instance network context. Should cover VPC config, VPC gateway configs, security group configs, NACLs, routing tables, firewalls for both the runner network location and the target GitLab Instance network location.
    - Use Windows 2019 instead of 1903.
    - Retry installations for AWS CLI for MSI error 1618 (MSI is processing another package).
    - Reboot behavior changed to support just one reboot while in launching lifecycle hook - to simplify idempotency checks in spin up automation.
    - Fix for Windows spot draining code.
    - Fail immediately if instance configuration script has non-zero exit. (#25)
    - Fixed "known problem: Windows machines are not completing autoscaling." noted in release v1.4.1-alpha7.
  • v1.4.3-alpha9
    Enable a much better form based experience without oddly named parameters using AWS::CloudFormation::Interface (#23)
  • v1.4.2-alpha8
    dfd800f8 · Pass 4ASGSpecifyVPC ·
    Enable selection of VPC.
  • v1.4.2-alpha7
    added easy button for linux docker single instance warm HA with scheduling ability
  • v1.4.1-alpha7
    v1.4.1-alpha7
  • v1.4.0-alpha6
    Support for arm64 architecture for Amazon Linux 2
  • v1.4.0-alpha5
    Simplification, implementation of s3 cache, list of tokens for linux
  • v1.4.0-alpha4
    83eb2a03 · Couple Doc updats ·
    First MVC Release - will need everyone's help to refine
  • v1.4.0-alpha3
    0aeb56e5 · v1.4.0-alpha3 ·
    First MVC ready for external testing
  • v1.4.0-alpha2
    72f16698 · fixes ·
    Ready for testing. Small updates since alpha1.
  • v1.4.0-alpha1
    764393f3 · Update ·
    Ready for testing by others
  • v1.3.0
    18c0ba5c · updates ·
    Works with Amazon Linux 2 and docker.  No memory metrics yet.