Skip to content

Latest commit

 

History

History
193 lines (129 loc) · 6.05 KB

README.md

File metadata and controls

193 lines (129 loc) · 6.05 KB

Universal Blue Update

Small update program written in python intended for use in Universal Blue that uses topgrade for executing updates.

Includes systemd timers and services for auto update

dependencies (fedora): sudo dnf install python3-psutil libnotify && pip install topgrade

Usage

Installation

You can add this to your image by simply pulling down and installing the rpm:

COPY --from=ghcr.io/ublue-os/ublue-update:latest /rpms/ublue-update.noarch.rpm /tmp/rpms/
RUN pip install topgrade && rpm-ostree install /tmp/rpms/ublue-update.noarch.rpm

If you are on an image derived from uBlue main:

COPY --from=ghcr.io/ublue-os/ublue-update:latest /rpms/ublue-update.noarch.rpm /tmp/rpms/
RUN pip install topgrade && rpm-ostree override remove ublue-os-update-services && rpm-ostree install /tmp/rpms/ublue-update.noarch.rpm

Note If you are on an image derived from uBlue main, you will need to remove or disable automatic updates with rpm-ostreed, to do this, you need to remove or change this line in the config file: AutomaticUpdatePolicy=stage (set to none if you don't want to remove the line)

Command Line

To run a complete system update, it's recommended to use systemd:

$ systemctl start ublue-update.service

This allows for passwordless system updates (user must be in wheel group)

Run updates from command line (not recommended)

Only run user updates (rootless):

$ ublue-update

Only run system updates (requires root):

$ pkexec ublue-update --system
usage: ublue-update [-h] [-f] [-c] [-u] [-w] [--system]

options:
  -h, --help         show this help message and exit
  -f, --force        force manual update, skipping update checks
  -c, --check        run update checks and exit
  -u, --updatecheck  check for updates and exit
  -w, --wait         wait for transactions to complete and exit
  --config CONFIG    use the specified config file
  --system           only run system updates (requires root)

Troubleshooting

You can check the ublue-update logs by running this command:

$ journalctl -exu 'ublue-update.service'

Configuration

Update Scripts

Update scripts are separated into two files inside of /usr/share/ublue-update

/usr/share/ublue-update/topgrade-system.toml

Topgrade config ran as root, handles rpm-ostree and flatpaks.

/usr/share/ublue-update/topgrade-user.toml

Topgrade config ran as user, handles flatpaks and distrobox containers.

See topgrade's GitHub for configuring these files.

Location

Valid config paths (in order of priority from highest to lowest):

  1. /etc/ublue-update/ublue-update.toml

  2. /usr/etc/ublue-update/ublue-update.toml

Config Variables

Section: checks

  • min_battery_percent: checks if battery is above specified percent

  • max_cpu_load_percent: checks if cpu average load is under specified percent

  • max_mem_percent: checks if memory usage is below specified percent

  • network_not_metered: if true, checks if the current network connection is not marked as metered

Section: checks.scripts

In addition to the predefined checks above, it is also possible to implement custom conditions through user-provided scripts and their exit codes. Each entry in the checks.scripts array must specify the following settings:

  • shell: specifies the shell used to execute the custom script (e.g. bash)

  • run: specifies the script text to be run using the specified shell

  • message: an optional message that is shown when the check fails

  • name: an optional human-readable name for this check

The parameters run and file are mutually exclusive, but at least one must be specified. The shell parameter is mandatory when using run.

The custom script should use its exit code to indicate whether the updater should proceed (exit code = 0) or whether updates should be inhibited right now (any non-0 exit code). If message is not specified but the script has written text to stdout, that text will be used as the message.

Section: notify

  • dbus_notify: enable graphical notifications via dbus

Full Example

[checks]
    min_battery_percent = 20.0  # Battery Level >= 20%?
    max_cpu_load_percent = 50.0 #     CPU Usage <= 50%?
    max_mem_percent = 90.0      #     RAM Usage <= 90%?
    network_not_metered = true  # Abort if network connection is metered

    [[checks.scripts]]
        name = "Example script that always fails"
        shell = "bash"
        run = "exit 1"
        message = "Failure message - this message will always appear"

    [[checks.scripts]]
        name = "Example script that always succeeds"
        shell = "bash"
        run = "exit 0"
        message = "Failure message - this message will never appear"

    [[checks.scripts]]
        name = "Example multiline script with custom message"
        shell = "bash"
        run = """
echo "This is a custom message"
exit 1
"""

    [[checks.scripts]]
        name = "Python script"
        shell = "python3"
        run = """
print("Python also works when installed")
exit(1)
"""

    [[checks.scripts]]
        name = "Example external script"
        # shell = "bash" # specifying a shell is optional for external scripts/programs
        file = "/bin/true"

[notify]
    dbus_notify = false         # Do not show notifications

How do I build this?

You can build and test this package in a container by using the provided container file.

  1. make builder-image will create a container image with all dependencies installed
  2. make builder-exec will execute a shell inside the builder container to allow you easily build the rpm package with make build-rpm
  3. make will trigger the build process and generate a .whl package that can be installed
  4. pip install --user -e . will allow to install an editable version of this package so you quickly edit and test the program

Special Thanks

Special thanks to cukmekerb for helping troubleshoot/add features early in project development