cornexo / clu-gopherbot

Configuration for Clu, my Gopherbot devel robot

Clu

Clu is the Gopherbot development robot that runs wherever I need him, and tracks with ongoing development of Gopherbot. This repository represents the bulk of Clu's custom configuration; the only other bits:

• clu-state - Clu's memories and other personal belongings; Clu's automated backup job updates this repository whenever his memories change
• clu-private (private repository) - Clu's private/environment - an environment file, which should be mode 0600 and closely guarded (see below for example contents)

Clu's custom configuration will form the basis for new robot configuration, the contents of robot.skel. At times, Clu's configuration may be ahead of master, when it represents a development target.

Directory Structure

This is the directory structure of a well-formed robot, created when I run ../gopherbot/fetch-robot.sh clu:

• clu/ - Top-level directory, not a git repo
  • gopherbot - symlink to /opt/gopherbot/gopherbot
  • terminal.sh - convenience script for running clu with the terminal connector
  • custom/ - clone of parsley42/clu-gopherbot, custom configuration repository
  • state/ - clone of parsley42/clu-state, where clu saves state, e.g. memories
  • private/ - clone of parsley42/clu-private, a private repository containing only environment
  • Makefile - a set of targets for running clu in a local container; see Running Clu
  • environment - a symlink to private/environment or .env if fetched from a .env file; see Fetching Clu

The environment file is only:

GOPHER_ENCRYPTION_KEY=<redacted>
#GOPHER_CUSTOM_REPOSITORY=https://github.com/parsley42/clu-gopherbot.git
GOPHER_CUSTOM_REPOSITORY=[e-mail]:parsley42/clu-gopherbot.git
DEPLOY_KEY=-----BEGIN=OPENSSH=PRIVATE=KEY-----:b3BlbnNzaC1rZXktdjEAAAAABG5vbmUAAAAEbm9uZQAAAAAAAAABAAABlwAAAAdzc2gtcn:NhAAAAAwEAAQAAAYEA7nXaHD1uZ9cZbI/szT74uOUwDTlihejMKKGPRcTRXbnnFkdLlcg+:fIJ09q8oIt33xvUQmmupnK65dqasQfna0jQeTFiJSdiXOjtH1kFmwiN7VyzrGu3Y8Yk2/i:tJ3By5TqsuqkoRDPxDGE6RnVrYYK3XTiVGPTNskkXctsJ2Ip9/W1rP5MY8oACgh1B8R24+:bXLwU0XzWDlrcVM9/l78AcSUEo6277cbBEwetVO5rR6dZJw+k/9Kfgyv+J4OucEXRMjgyZ:k6gz1P+TePCAntw6y9ce+UBFdXeBavHom3Pc1k5o6ywB53UHmefYiyJ0YteCfnfaQAs+Gt:ra6AW8nGosVJegPapDtT/lAKHCH0JwFp8Xj6ksMTAkWao28U62F0Amtk3a1Fm/u7ZAn5CB:1FdTn5mcwv6dYDIo6NVVeHDZvpEk7kl4q9QV8bEVtYkNMN7dHDXZE12+IFSIbfS1DvuwCf:k2cMaxMFY4tPricejk/2oTg/mPAt+pG+KPh6Gi69AAAFkEMibBpDImwaAAAAB3NzaC1yc2:EAAAGBAO512hw9bmfXGWyP7M0++LjlMA05YoXozCihj0XE0V255xZHS5XIPnyCdPavKCLd:98b1EJprqZyuuXamrEH52tI0HkxYiUnYlzo7R9ZBZsIje1cs6xrt2PGJNv4rSdwcuU6rLq:pKEQz8QxhOkZ1a2GCt104lRj0zbJJF3LbCdiKff1taz+TGPKAAoIdQfEduPm1y8FNF81g5:a3FTPf5e/AHElBKOtu+3GwRMHrVTua0enWScPpP/Sn4Mr/ieDrnBF0TI4MmZOoM9T/k3jw:gJ7cOsvXHvlARXV3gWrx6Jtz3NZOaOssAed1B5nn2IsidGLXgn532kALPhra2ugFvJxqLF:SXoD2qQ7U/5QChwh9CcBafF4+pLDEwJFmqNvFOthdAJrZN2tRZv7u2QJ+QgdRXU5+ZnML+:nWAyKOjVVXhw2b6RJO5JeKvUFfGxFbWJDTDe3Rw12RNdviBUiG30tQ77sAn5NnDGsTBWOL:T64nHo5P9qE4P5jwLfqRvij4ehouvQAAAAMBAAEAAAGBALD0QbO9HoXuSA6YyzgP59CFOu:BFWkhW1dG8+i3i/R7ZSpPsujlfTIdm49b/agBdyXYZ+4UsKcR8oGJdEu0utWRRir5K4S4s:jSSIQynKhK/CVs/9JEZqhBfRJD7+7qNpqVWokEuMBRUmyb9q5oHnnTQ5LNHvtSzLUWFGeK:AitDnDNGYdgLKbLPfrHzTq1B7Jv4fGyHJzMT6h9Yo2JIX0BHxnXR5cS4Kd1W2d8xfKFrpS:QqgbjhCTXLsnPRp4aCMOFoBqogqRFrUL7XhgcYxSim2rpRc5DhtUMN/rMi/iYZ2ECSA8kP:Si3gJEaWutwdAWtPvM5XVEbUMRo6o7/h9XZmeiTODF5aGz24PWelJQB76eZpr/jUubvS8w:94g1xaKKu1ra/9KmPRV2ouIiJy+j7Bql5XorULfT0HtYrTgMRCd9pvBB9VRZPCVJeJ0drX:0pKkc+OG69ptWmNnDB/f31vqapqZNDrm/jeeFUXIMp/aSHDxu58RqNnFr8V7QgC4iT/QAA:AMEArfJ++TXOiQoxJW2ZSTtU7uetPR+aC/enYvm2fVu59cy74UvU8Mwbz5FTwF48llI7u9:bUfEQxYf4S7kvoH9GcQAiiG3rOBDjbBL9jRjWsrxPIAlRygifDiE3KdGxwekfZxuW2kV2O:t2yz3L00XS6yV/+pvIIqo1XiWCuXZj7EaSqEkPTb7l20G7qctkqRNwFxkJd0msF+c/Uodl:I7WfN9AXQ2INUF8MAIAfaTwgfHtd/7pRGTB+96FzR7CcqhOdJFAAAAwQD/niUQ+GWTJrWu:yp46hQ1JHOAgtP1vFLYjXLdTnevPADQrbe5HX8Jmw1RbgAfBociIyTjATvXmtCBhyjGZLE:em6WdUlL8dw+XU6DYUdPiIb957J3wJwmtaecQlknxr8O/emeiMaGn2Orl1b6CNGjUM5O5m:tceOSA5u9MqlxVPGqZFKHc+dEJZkofXoxLOwdhTE8EQmChkmbtQyOLh+IvOfr4bSz18ALY:j00u+YBmf+DogCegDTpzqMxsAGmeSZw2sAAADBAO7RI5fOmviMoKMs9UicJ7w6myvwQo7z:2XA+7bpGCV5b4D0vZ8iYHmlgJwsPqf4Wcnid1nHGSYt0BXjpH0yK9+gD+H28xOo3E+s1Co:ieZXJvYtPyX7ASN2cfe6FZc4nH+2ACYkH2EFiaKc1ZvsPGRcEbrqTPbKGVNTnNpjGWVjBO:vPcoLByZo1BzxvfdNtRxNfH7GNmI7Neo75PufeNQ1tdfr9/b2hmVzrFWzIk/OUudnvhkgx:qsJQRfxLUub88IdwAAABhwYXJzZUBoYWt1aW4ubG9jYWxkb21haW4B:-----END=OPENSSH=PRIVATE=KEY-----:

NOTE: You can use 'tr' to create the above DEPLOY_KEY=..., e.g.:

$ cat deploy_rsa | tr ' \n' '=:'
-----BEGIN=OPENSSH=PRIVATE=KEY-----:b3Bl...W4B:-----END=OPENSSH=PRIVATE=KEY-----:

Fetching Clu

The fetch-robot.sh script will create the above directory structure. There are two ways I can run gopherbot/fetch-robot.sh to fetch Clu:

Fetching with .env

If I create an empty directory for Clu and stick the above environment in a .env file in that directory, I can just:

$ cd clu
$ /opt/gopherbot/fetch-robot.sh

... that will read the .env and create the directory structure above. This will work for any well-formed robot.

Fetching during development

Because I do most of my development in a github.com/parsley42/gopherbot fork, and Clu's repositories are also parented by github.com/parsley42, I can fetch Clu with the developer shortcut mode of fetch-robot.sh:

$ ./gopherbot/fetch-robot.sh clu

This requires not only that it be run from a forked Gopherbot repository, but that the development robot has a private repository (e.g. clu-private) with it's environment file (containing the required GOPHER_ENCRYPTION_KEY). It's definitely the fastest way for me to get up and going, though.

Running Clu

One of the main goals for Gopherbot version 2 was robot portability and first-class container support. Given the new feature set, there are several ways I can start Clu.

Worth noting here is how Gopherbot get's it's environment. The first and most obvious is from the environment exported from the parent process, as is the case when starting in a container. Second, on startup, Gopherbot first looks for private/environment, then .env, and loads the first one if finds, overwriting the current environment.

I can start that robot with TWO environment variables

(... for anybody who remembers the old "Name That Tune" game show)

Since Clu's primary configuration repository is publicly available, I can start in a directory with this .env file:

GOPHER_ENCRYPTION_KEY=<redacted>
GOPHER_CUSTOM_REPOSITORY=https://github.com/parsley42/clu-gopherbot.git

Note that this is closer to how I would run Clu in a production container or vm, without using fetch-robot.sh, though I would likely use a deploy key as shown above in production.

Method 1 Local Install

$ /opt/gopherbot/gopherbot 
2020/01/06 11:53:02 Initialized logging ...
... (clone, restart, and restore brain) ...
2020/01/06 11:53:07 Info: Robot is initialized and running

After which you'll have a basic directory structure with GOPHER_CUSTOM_REPOSITORY in custom/, and GOPHER_STATE_REPOSITORY (defined in custom/conf/gopherbot.yaml) restored in to state/.

Method 2 Run in container with Docker

$ docker run --env-file .env lnxjedi/gopherbot:latest
2020/01/06 11:53:02 Initialized logging ...
... (clone, restart, and restore brain) ...
2020/01/06 11:53:07 Info: Robot is initialized and running

This accomplishes the same as above, only custom/ and state/ are created in the container's /home/robot directory.

Using fetch-robot

Most of the time when I'm doing development, I just use fetch-robot.sh like so:

$ ./gopherbot/fetch-robot.sh clu
$ cd clu
$ ./gopherbot
2020/01/06 11:53:02 Initialized logging ...
... (normal startup, custom/ and state/ already present) ...
2020/01/06 11:53:07 Info: Robot is initialized and running

Since fetch-robot.sh also copies the sample container Makefile from gopherbot/resources/docker, I can also use e.g. make prod, make dev and make debug targets to start Clu