5.7 KiB
docker-gentoo-steemd
steemd built inside a Docker container using a Gentoo base with size optimization
In the dkr directory is a short shell script that sets some handy aliases, which you
will need to run this docker container:
source init.sh
I recommend on the servers you use this script that you put this on your shell startup
script, it will make it a lot simpler for you. While you are still in the dkr directory,
run this command:
echo "source `pwd`/init.sh" >> ~/.bashrc
Substitute .bashrc with .zshrc if you, like me, use zsh.
Note that you can 'source' this file from any other part of the filesystem and it will still know where its' parent directory and all the bits are.
Then you can type halp and it will show you the short commands to perform all the
functions. The script stores its own location automatically so all commands will work
no matter where you move in your filesystem. The script can be edited with the command:
.editsh. Note that the halp command simply processes the script itself to produce
the output, the code is the documentation, is the code.
Basic info about managing a Docker Witness instance
Because it can be a bit confusing (it confused me for a long time) I will explain the
basic procedure to using this, beyond what the help in the init.sh provides:
To build the image, .build will run the procedure defined in the Dockerfile. Then
there will be an image you can see with dkr ps -a (probably at the top of the output).
To get the image running, type .run (this will also start the steemd which, with an empty data
folder will begin syncing from scratch, and use an empty configuration). There is
a file in data/witness_node_data_dir/ called config.ini.example which will help
you create a basic witness configuration, I have annotated it with comments to
explain what you need to put in.
It is confusing, but .run is very different to .start. The former initialises
a newly built container and starts it up, the latter starts an already initialised
container again after you have stopped it. It will complain if you try to .run
after .run has already been issued, and will only work if you .stop and .rm
to remove the container. Then you will have to .build again. This is what you
want to do if you edit the Dockerfile.
.stop will stop a running container, .start will start it again. .log will
show you the log file output, which is created in data/steemd.sh and is there
so that other scripts can monitor and parse the log file. If you use the
sudo docker <image name> logs file instead, you will see a notice that the
logs are going to a log file and how to view them as they come down.
.enter will let you run a shell inside the container and poke around if you want
to do that (any changes will be lost on a .stop;.rm; though the contents of
/work are persistent outside of the container.
The advantage of this arrangement is that even if you delete the Docker container (not this folder), you won't lose the block_log and configuration files.
The data directory contains a volume that is mounted inside the container as /work
which I have done so you can easily import block_log files or even the whole
blockchain folder inside witness_node_data_dir. However, note that this by default
will build v0.17.1 so an older shared_memory.* file will not work.
Lastly, in my init.sh script you also have two commands, one, .editsh lets you
edit the init.sh file (after you first run it from the dkr/ directory it will
remember even if you edit it where it is) and the other is .editdkr which will
let you edit the Dockerfile.
There is more commands now available in there, but the 'halp' command will provide enough information to use most of them.
There is now a monitor script that can be used, and there is a command in the init.sh
that launches it inside a screen session detached. Copy or rename the config.example
and config.py.example to the same names without the .example ending and the monitor
script will have all the details. Some of the parameters are unneccessary in the current
configuration but I intend to add more functions that will use them (such as a manual
toggle between your primary and secondary witness).
There is also a price feed setter script that draws its numbers from coinmarketcap,
it also draws on the the same configurations, and note that the init.sh brings the
environment variables set in config into your environment along with all the aliases
to make managing a witness is nice and easy.
This is a super simple, minimalistic docker container, designed in the way that I do things, which is usually different to other people. I think it is simpler and more efficient.
The configuration of the Gentoo system inside the container is built with -Os
compiler optimisation flag enabled, so that most of the support libraries are probably
smaller and faster than the ones that the host system would use. Neither Boost nor
steemd is optimised this way, however. I may amend this in the future, but for
now, this container is fully working and has a side benefit that following the
commands inside the Dockerfile you can also build steemd to run on a Gentoo server.
Prerequisites
The accessory scripts in this package assume a number of prerequisites, which are mostly generally already present. The exceptions are the python related stuff (not necessary for the basic witness operation but for the feeder and monitor they are required)
For Ubuntu to install these prerequisites:
sudo apt install python3 python3-pip screen
sudo -H pip3 install asyncio requests piston-lib websockets
For other distros, like Arch linux, the information will be added in the near future.