If you own Tezos tokens (XTZ), you want them to work for the Tezos network and ecosystem. It is a fun challenge and you will be rewarded for doing it. The more individual bakers we have in the Tezos ecosystem the more decentralized and resilient Tezos will become.
Granted, it takes a little bit of work - but good guides exist and I think you will find it worth while. This repository focuses on getting you maximal uptime and ease of use once you have installed your node(s) and gotten your ledger to work.
To spin up a node use this excellent guide: http://doc.tzalpha.net/introduction/howtoget.html#build-from-sources
To get your Ledger Nano S to work with Tezos follow this excellent guide: https://github.com/obsidiansystems/ledger-app-tezos/blob/master/README.md
If you are not familiar with "services" or systemd there is a good intro here: https://www.digitalocean.com/community/tutorials/how-to-use-systemctl-to-manage-systemd-services-and-units
Briefly, what we want is a stable system with maximal uptime in return for minimal intervention. As such we need a tezos node that starts itself once the system boots and a baker/endorser/accuser that is always on and ready to bake/endorse/accuse. systemd can easily help us achieve this.
Below you'll find the config files for such a system together with some explanation of how it works.
Basically, for a full bakery we want to configure and (auto-)run four services
- Tezos Node
- Tezos Baker
- Tezos Endorser
- Tezos Accuser
For a node only, the bottom of this page has instructions on how to use systemd to run a non-baking node. It is super easy. If you only need that then you can read the tezos-node.service part and then skip to the bottom of the document.
The individual files for a full bakery are outlined below. You can copy and paste them into the paths/files mentioned in each section.
For monitoring the daemons you can use journalctl which is a powerful tool to monitor services both real-time and after the fact.
There is separate section on this towards the end of this document; go there now by clicking this link
First we want to set up the service to run our Tezos node. The below example includes running over VPN - you dont have to, just remove the openvpn-client@<vpnprovider>.service under the [Unit] both for Wants and After.
At the ExecStart line you can place whatever command you normally use to start your node - just dont use nohub etc. You can (& should) change your userand group to whatever you use on your system. I control my nodes in a simple config.json - but the below should work with default settings all over.
# The Tezos Node service (part of systemd)
# file: /etc/systemd/system/tezos-node.service
[Unit]
Description = Tezos Node Service
Documentation = http://tezos.gitlab.io/betanet/
Wants = network-online.target openvpn-client@<vpnprovider>.service
After = network-online.target openvpn-client@<vpnprovider>.service
[Service]
User = baker
Group = baker
WorkingDirectory= /home/baker/
ExecStart = /home/baker/tezos/tezos-node run --bootstrap-threshold=1
Restart = on-failure
[Install]
WantedBy = multi-user.target
RequiredBy = tezos-baker.service tezos-endorser.service tezos-accuser.service
To install your tezos node as a service that loads at boot time you use systemctl:
sudo systemctl enable tezos-node.service
Notice that we use After to tell systemd that this service should be loaded after networking is established. We allow the tezos-node service to load without network by using Wants as we might establish network later. You can make sure the tezos node would only load once network is established by chainging 'Wants' to 'Requires' but I would recommend against doing so, as it limits your flexibility. Same with the VPN service
To start your tezos node service you would use:
sudo systemctl start tezos-node.service
Obviously, you rarely use this command unless you 1) havent rebooted your system after installing the service or 2) you have actively shut down the service
To start your tezos node service I strongly recommend you use the reload-or-restart instead, as this would ensure (via systemd) that the service is started if not running already and reloaded if already running:
sudo systemctl reload-or-restart tezos-node.service
To see the status of your tezos node service you would use:
sudo systemctl status tezos-node.service
To stop your tezos node service you would use:
sudo systemctl stop tezos-node.service
If for some reason you don't want the node to start at boot anymore, simply do:
sudo systemctl disable tezos-node.service
Now that the node is up and running we want to run the baker, endorser and accuser the same way.
# The Tezos Baker service (part of systemd)
# file: /etc/systemd/system/tezos-baker.service
[Unit]
Description = Tezos Baker Service
Wants = network-online.target openvpn-client@<vpnprovider>.service
BindsTo = tezos-node.service
After = tezos-node.service
[Service]
User = baker
Group = baker
WorkingDirectory= /home/baker/
ExecStartPre = /bin/sleep 1
ExecStart = /home/baker/tezos/tezos-baker-004-Pt24m4xi run with local node /home/baker/.tezos-node ledger_bakerone_ed_0_0
Restart = on-failure
[Install]
WantedBy = multi-user.target
We know that these services require a tezos node and therefore we require the node to be running first - the hardest form of requirement is binding - this means that this service will only start if the service it BindsTo is successfully started and running. Also, if the service (Tezos Node) this service (Tezos Baker) binds to crashes this service will be stopped.
You should replace the ExecStart command with whatever command you want to run your baker with. Also, replace the ledger_bakerone_ed_0_0 with whatever alias your baking key has. Notice the ExecStartPre: It is a little hackish, but I found that introducing a one second delay between starting the node and the baker, endorser and accuser would make the service run smoothly. Else systemd will start them too closely together. There are ways to adjust this using sytemd, but to keep things simple, we simply sleep for a second prior to executing the command to fire up the baker.
We also know that it would probably be good to reload the baker, endorser and accuser should the node ever reload and therefore we use BindsTo to bind these services to the node. This effectively means, that all four services will restart if you restart the node and that you can restart each of the baker, endorser and accuser services seperately, should you need to.
Same commands as for node to enable, reload/start, stop and get status on the baker:
sudo systemctl enable tezos-baker.servicesudo systemctl reload-or-restart tezos-baker.servicesudo systemctl stop tezos-baker.servicesudo systemctl status tezos-baker.service
Note: You must have the Tezos baking app open on your Ledger Nano S when you (re)start your baker and endorser.
Now we simply do the same with the endorser and accuser daemons.
# The Tezos Endorser service (part of systemd)
# file: /etc/systemd/system/tezos-endorser.service
[Unit]
Description = Tezos Endorser Service
Wants = network-online.target openvpn-client@<vpnprovider>.service
BindsTo = tezos-node.service
After = tezos-node.service
[Service]
User = baker
Group = baker
WorkingDirectory= /home/baker/
ExecStartPre = /bin/sleep 1
ExecStart = /home/baker/tezos/tezos-endorser-004-Pt24m4xi run ledger_bakerone_ed_0_0
Restart = on-failure
[Install]
WantedBy = multi-user.target
Same commands as for node to enable, reload/start, stop and get status on the baker:
sudo systemctl enable tezos-endorser.servicesudo systemctl reload-or-restart tezos-endorser.servicesudo systemctl stop tezos-endorser.servicesudo systemctl status tezos-endorser.service
Note: You must have the Tezos baking app open on your Ledger Nano S when you (re)start your baker and endorser.
# The Tezos Accuser service (part of systemd)
# file: /etc/systemd/system/tezos-accuser.service
[Unit]
Description = Tezos Accuser Service
Wants = network-online.target openvpn-client@<vpnprovider>.service
BindsTo = tezos-node.service
After = tezos-node.service
[Service]
User = baker
Group = baker
WorkingDirectory= /home/baker/
ExecStartPre = /bin/sleep 1
ExecStart = /home/baker/tezos/tezos-accuser-004-Pt24m4xi run
Restart = on-failure
[Install]
WantedBy = multi-user.target
Same commands as for node to enable, reload/start, stop and get status on the accuser:
sudo systemctl enable tezos-accuser.servicesudo systemctl reload-or-restart tezos-accuser.servicesudo systemctl stop tezos-accuser.servicesudo systemctl status tezos-accuser.service
sudo systemctl status tezos-node.service tezos-baker.service tezos-endorser.service tezos-accuser.service
Or alternatively, shorter but less ordered:
sudo systemctl status 'tezos-*.service'
You can restart all four services by restarting the node (because we bound the baker/endorser/accuser to the node):
sudo systemctl reload-or-restart tezos-node.service
Similarly all services will stop upon:
sudo systemctl stop tezos-node.service
And you can stop the baker/endorser/accuser individually if you want.
sudo systemctl stop tezos-baker.servicesudo systemctl stop tezos-endorser.servicesudo systemctl stop tezos-accuser.service
The above configurations work for both front end nodes and for baking/endorsing/accusing nodes. If you want to use this (I recommend you do) for your front-end nodes too, simply remove the line RequiredBy = tezos-baker.service tezos-endorser.service tezos-accuser.service from your tezos-node.service and do not install the baker/accuser/endorser services.
Above, we use Restart:on-failure. You could use Restart:always - I just haven't found it neccessary and there is a slight risk that - if combinded with loose restart settings - could exhaust your system. But feel free to try it out if it'll make you sleep better at night. Now that you have all your Tezos operations 'servicified' you can indeed start sleeping at night again, without loosing out on your baking and endorsing slots.
Just include the following in your [Service] section
Environment = TEZOS_LOG="client.signer.ledger -> debug"
If you need to pass a lot of environment variables, use EnvironmentFile instead and place one variable per line here. EnvironmentFile should point to your file, e.g. /home/baker/tezosenvironmentvariables.
Sometimes it is neccessary to go through log files to identify root causes for different events and sometimes it is just fun to follow the your Tezos services (=daemons) live. systemd has a very powerful tool to do this - it is called journalctl and a few examples on how to use it are given below.
To simply follow your node's output real-time:
journalctl --follow --unit=tezos-node.service
You dont really need to add the .service - but I'll keep doing it here for clarity
Similarly with the baker, endorser and accuser:
journalctl --follow --unit=tezos-baker.servicejournalctl --follow --unit=tezos-endorser.servicejournalctl --follow --unit=tezos-accuser.service
You can also get the output formatted to suit your needs. Try for example:
journalctl --follow --unit=tezos-endorser.service --output=json-pretty
Tezos runs its time by the universal timezone 'UTC' to get journalctl to output your log in utc simply add --utc:
journalctl --follow --unit=tezos-endorser.service --utc
By now you've probably understood that the possibilities are almost endless and the flexibility is second to none. Try for example to get your log for the endorser after a given timestamp or between two timestamps by doing these:
journalctl --unit=tezos-endorser.service --since=yesterdayjournalctl --unit=tezos-endorser.service --since=todayjournalctl --unit=tezos-endorser.service --since='2018-08-01 00:00:00' --until='2018-08-10 12:00:00'
Or find your bakes since last boot:
journalctl --unit=tezos-baker.service --boot=-0 | grep candidate
If you have installed/compiled journalctl with pattern matching functionality you can do:
journalctl --unit=tezos-baker.service --boot=-0 --grep=candidate
And on and on....
See more using man journalctl
I've now been asked repeatedly for a donation address. Donations are not expected. If you feel you want to anyway you can use: tz1a2oGa6yTXGuS9d9DTckQm5vrh12qYqCqL
Enjoy!