v2.0.0

The latest version of jailmaker is now officially licensed as open source under the LGPL-3.0 license! This makes it possible for iXsystems to reuse jailmaker code for a future release of TrueNAS SCALE. This version includes breaking changes, so please read these notes carefully. You can, as usual, replace jlmkr.py to upgrade to the latest version.

Removed config options

The deprecated gpu_passthrough and docker_compatible config values have been removed. Please check your config files and remove these options if you're still using them. Instead of gpu_passthrough use the dedicated gpu_passthrough_intel and gpu_passthrough_nvidia options. Instead of docker_compatible refer to the documentation on how to make a docker-compatible jail.

Removed install command

The jlmkr install command has been removed. This doesn't change much for existing users. The shell alias which jailmaker created automatically in /root/.zshrc and /root/.bashrc will continue to work. However I suggest you remove the # managed by jailmaker comment from these files, as jailmaker will no longer manage this alias and won't modify your shell aliases from now on. The documentation explains how to manually setup a new alias if desired.

What's Changed

• v2.0.0 by @Jip-Hop in #187
• Removed deprecated config options
• Removed the jlmkr install command
• Added license and copyright info
• Removed tab auto complete from interactive create

Full Changelog: v1.5.0...v2.0.0

v1.5.0

There should be no breaking changes. You can simply replace jlmkr.py to upgrade to the latest version.

What's Changed

• Add post_start_hook to support multiple network bridges by @Jip-Hop in #185
• Add documentation for multiple network bridges by @RyanMelena in #182

New Contributors

• @RyanMelena made their first contribution in #182

Full Changelog: v1.4.2...v1.5.0

v1.4.2

This bugfix release fixes issue #177 where a zfs pool with a space in it would cause problems when running the pre_start_hook and post_stop_hook. Creating the jlmkr alias will now also work if the zfs pool contains a space. There should be no breaking changes. You can simply replace jlmkr.py to upgrade to the latest version.

What's Changed

• seccomp documentation by @izenn in #171
• Properly handle path with spaces by @Jip-Hop in #180

New Contributors

• @izenn made their first contribution in #171

Full Changelog: v1.4.1...v1.4.2

v1.4.1

This bugfix release fixes issue #127 where nvidia passthrough would not work properly under some conditions. There should be no breaking changes. You can simply replace jlmkr.py to upgrade to the latest version.

What's Changed

• Fix for Nvidia Passthrough by @dalgibbard in #166

New Contributors

• @dalgibbard made their first contribution in #166

Full Changelog: v1.4.0...v1.4.1

v1.4.0

This release brings improved behavior of the initial_setup script when creating new jails from config templates. This release does not introduce any changes in how existing jails are treated. In general there should be no breaking changes. But if you have your own, custom, config templates you may want to read carefully if this release impacts your initial_setup script. You can simply replace jlmkr.py to upgrade to the latest version.

What's Changed

• The initial_setup script will now run once the jail has been startup for the first time and will wait for the init system and networking to be ready. It used to run BEFORE the first start using the host networking namespace and without applying the systemd_nspawn_user_args. This had several downsides, such as the inability to install some packages, not being able to access bind mounts from the initial_setup script and inconsistent networking modes. The new behavior fixes this.
• Config templates now default to bridge networking, please see the updated docs on how to set up bridge networking.
• The docker config template will now also install the NVIDIA Container Toolkit if gpu_passthrough_nvidia=1 when the jail starts for the first time.
• Updated experimental lxd and incus templates to take advantage of the improved initial_setup behavior.
• Fixes in the podman rootless install instructions as well as the config template.

Full Changelog: v1.3.0...v1.4.0

v1.3.0

There should be no breaking changes. You can simply replace jlmkr.py to upgrade to the latest version.

What's Changed

• Follow Debian guidelines for launching editors by @maxnordlund in #157

New Contributors

• @maxnordlund made their first contribution in #157

Full Changelog: v1.2.1...v1.3.0

v1.2.1

This release contains a bugfix to support zfs pools and datasets containing a space in their name. There should be no breaking changes. You can simply replace jlmkr.py to upgrade to the latest version.

What's Changed

• Update README.md by @Jip-Hop in #136
• add .gitignore by @mrstux in #126
• Support spaces in pool and dataset name by @Jip-Hop in #141

New Contributors

• @mrstux made their first contribution in #126

Full Changelog: v1.2.0...v1.2.1

v1.2.0

This release combines the changes previously released as v1.1.4 and v1.1.5. Because those releases didn't follow use semantic versioning, releases v1.1.4 and v1.1.5 will be removed to correct this mistake. If you are already on version v1.1.5 then this release only changes the version number to v1.2.0. If you're on v1.1.4 or before you may want to read these release notes carefully as new behavior is introduced and deprecated features are announced.

This release includes the feature request to create a ZFS dataset for each jail thanks to @templehasfallen. This behavior is now the default for new users. Existing users may want to migrate their existing setup, which uses plain directories for their jails, by following the migration guide.

This release deprecates the config options gpu_passthrough and docker_compatible. They will be removed in a future release. To be prepared I recommend to no longer use these deprecated config options when creating new jails and replace these options in existing jails which still use them.

• The gpu_passthrough option should be replaced with gpu_passthrough_intel and/or gpu_passthrough_nvidia
• The docker_compatible is not actually required to run docker containers in a jail. You may look at the docker config template to see how to run docker in a jail.

Additionally this release introduces the new seccomp config option which allows you to completely turn off seccomp filtering. This may increase performance at the expense of security. When seccomp is off you don't need allow syscalls with --system-call-filter= (no syscalls will be blocked by seccomp). This behavior was previously included with the docker_compatible option, but is now separated in its own dedicated option.

There should be no breaking changes. You can simply replace jlmkr.py to upgrade to the latest version.

What's Changed

• Added Full ZFS Dataset Support by @templehasfallen in #118
• Deprecated docker_compatible and gpu_passthrough config options and added seccomp config option in #121
• Correct semver by @Jip-Hop in #134

New Contributors

• @templehasfallen made their first contribution in #118

Full Changelog: v1.1.3...v1.2.0

v1.1.3

What's Changed

• Hotfix for hostname of jail not set correctly by @mschabhuettl in #116

New Contributors

• @mschabhuettl made their first contribution in #116

Full Changelog: v1.1.2...v1.1.3

v1.1.2

This version uses the updated lxc-download script from upstream which contains fixes for column alignment and preserves xattrs on unpack. In addition to that this release contains improvements for error handling and brings back warnings to the non-interactive create process.

There should be no breaking changes. You can simply replace jlmkr.py to upgrade to the latest version.

What's Changed

• Format templates/incus/README.md benchmarks by @hugojosefson in #110
• Release 1.1.2 by @Jip-Hop in #114

New Contributors

• @hugojosefson made their first contribution in #110

Full Changelog: v1.1.1...v1.1.2