script to rest thin-edge.io state (but persist the device certificates)

README.md

tedge-reset

What does it do?

The script does the following things (but please read the script yourself before executing it):

• stop the main tedge services
• stop mosquitto
• delete the mosquitto database
• delete internal tedge state
• delete thin-edge.io file based logs and file-transfer cache

Afterwards, it will prompt the user if they wish to reconnect. It also directs the user to delete the device (and all child managed objects) from the cloud before reconnecting.

How to use it

Copy the tedge-reset.sh script to the device, and then run the following commands:

sudo chmod a+x ./tedge-reset.sh
sudo ./tedge-reset.sh

Deleting the device from the cloud

When deleting the device from the cloud, if you're using go-c8y-cli, then you MUST use the --cascade option to ensure that all of the child devices are also deleted, otherwise it can result in unlinked services and child devices.

Below shows an example of the correct usage to delete the device and all of its child devices and services.

c8y devices list --name "example01" | c8y devices delete --cascade

In the event that you have deleted the device managed object and not the child devices and services, then you will have to manually delete them using the following command where all inventory managed object which are owned by the device's user are searched for then deleted (note: that the owner has the device_ prefix):

WARNING: Please verify the find results before deleting it to ensure the query is returning the expected amount of inventory managed objects!

c8y inventory find --owner "device_example01" --includeAll | c8y inventory delete

tedge-reset.sh

#!/bin/sh
set -e

# allow for custom installation directories
TEDGE_CONFIG_DIR="${TEDGE_CONFIG_DIR:-/etc/tedge}"

stop_services() {
    tedge disconnect c8y
    systemctl stop tedge-agent
    systemctl stop mosquitto
    systemctl stop tedge-container-plugin >/dev/null 2>&1 ||:
}

delete_state() {
    # NOTE: This assumes the default mosquitto.db location, but
    # users can change this location, so it is up to the users to
    # remove the mosquitto.db themselves
    if [ -f /var/lib/mosquitto/mosquitto.db ]; then
        rm -f /var/lib/mosquitto/mosquitto.db
        echo "Deleted the /var/lib/mosquitto/mosquitto.db file"
    fi

    echo "Removing internal thin-edge.io state" >&2
    rm -rf "$TEDGE_CONFIG_DIR/.agent/"
    rm -rf "$TEDGE_CONFIG_DIR/.tedge-mapper-c8y/"

    if [ -n "$(find "$TEDGE_CONFIG_DIR/operations/c8y/" -type d -mindepth 1 -maxdepth 1 2>/dev/null ||:)" ]; then
        find "$TEDGE_CONFIG_DIR/operations/c8y/" -type d -mindepth 1 -maxdepth 1 -exec rm -rf {} \;
    fi

    echo "deleting logs" >&2
    LOGS_PATH=$(tedge config get logs.path)
    if [ -d "$LOGS_PATH" ]; then
        rm -f "$LOGS_PATH"/agent/*.log
    fi

    DATA_PATH=$(tedge config get data.path ||:)
    if [ -d "$DATA_PATH" ]; then
        rm -rf "$DATA_PATH"
    fi

    # optional: remove the device certificate, but if this were to be done
    # then the device would also have to be re-registered with Cumulocity
    # which is more involved
    #rm -f /etc/tedge/device-certs/*
}

post_start() {
    tedge connect c8y
    systemctl start tedge-container-plugin >/dev/null 2>&1 ||:
}

main() {
    stop_services

    delete_state

    # Prompt user if it is ok to start services again
    DEVICE_ID=$(tedge config get device.id ||:)
    printf "Now delete the item from the cloud (c8y_Serial=%s) and then type 'y': " "$DEVICE_ID" >&2
    read -r ANSWER

    case "$ANSWER" in
        y|Y|yes)
            echo "Connecting thin-edge.io again to Cumulocity" >&2
            post_start
            ;;
        *)
            echo "NOTE: thin-edge.io was not reconnected, please run 'tedge reconnect c8y' when you're ready" >&2
            ;;
    esac
}

main "$@"