Skip to content

Latest commit

 

History

History
82 lines (52 loc) · 4.3 KB

performance_considerations.md

File metadata and controls

82 lines (52 loc) · 4.3 KB

Performance considerations

Performance is often important when running GitHub action runners with garm. This document shows some ways to improve the creation time of a GitHub action runner.

GARM specific performance considerations

Bundle the GitHub action runner

When a new instance is created by garm, it usually downloads the latest available GitHub action runner binary, installs the requirements and starts it afterwards. This can be a time consuming task that quickly adds up when a lot of instances are created by garm throughout the day. Therefore it is recommended to include the GitHub action runner binary inside of the used image.

There are two ways to do that:

  1. Add the extracted runner to /opt/cache/actions-runner/latest in which case, garm won't do any version checking and will blindly trust that whatever you put there is indeed the latest. This is useful if you want to run a pre-release of the runner or have your own patches applied to it. Also GitHub runners have an auto-update mechanism. When it detects that a new version is available, it updates itself to the latest version.

  2. Add the extracted runner to /opt/cache/actions-runner/$VERSION where $VERSION is the version of the runner. In this case, if what garm fetches from GitHub is different than what you bundled in the image, it will download and install the version indicated by GitHub.

Note, when bundling the runner with your image, you will have to download it, extract it to one of the above mentioned locations and also run the ./bin/installdependencies.sh inside the extracted folder. All dependencies needed to run the runner must be pre-installed when bundling.

Example steps:

# Create a temporary instance from your base image
lxc launch <BASE_IMAGE> temp

# Enter bash inside the container
lxc exec temp -- bash

# Get and install the runner
mkdir -p /opt/cache/actions-runner/latest
cd /opt/cache/actions-runner/latest
curl -o actions-runner-linux-x64-2.305.0.tar.gz -L https://github.com/actions/runner/releases/download/v2.305.0/actions-runner-linux-x64-2.305.0.tar.gz
tar xzf ./actions-runner-linux-x64-2.305.0.tar.gz
./bin/installdependencies.sh 

# Exit the container
exit

# Stop the instance and publish it as a new image
lxc stop temp
lxc publish temp --alias BASE_IMAGE-2.305.0

# Delete the temporary instance
lxc delete temp

# Update garm to use the new image
garm-cli pool update <POOL_ID> \
  --image=BASE_IMAGE-2.305.0

Disable updates

By default garm configures the cloud-init process of a new instance to update packages on startup. To prevent this from happening (and therefore reduce the time needed to start an instance) garm can be configured accordingly.

Example to disable this on LXD provider:

garm-cli pool update <POOL_ID> \
  --extra-specs='{"disable_updates": true}'

LXD specific performance considerations

Storage driver

LXD supports various storage drivers out of the box. These storage drivers support different features which influence the creation time of a new instance. Most notably check if the driver supports Optimized image storage and Optimized instance creation as these have the biggest impact on instance creation time.

If you're not sure which storage driver is currently used, check your storages with lxc storage list.

Use shiftfs/idmapped mounts

Whenever a new unprivileged instance is started on LXD, its filesystem gets remapped. This is a time consuming task which depends on the image size that's being used. For large images this can easily take over a minute to complete. There are two ways to get around this: shiftfs or idmapped mounts. While the latter is the preferred one, not all filesystems currently support it, so in most cases enabling shiftfs show a significant performance improvement.

Example on how to enable it on a snap installed LXD:

snap set lxd shiftfs.enable=true
systemctl reload snap.lxd.daemon

Some details and discussions around shiftfs can be found here.

Note: When shiftfs is used, mounting between host and container might need some extra steps to be secure. See here for details.