Problems with Zrok Agent systemd service

zrok
1

I have a ubuntu server machine with zrok client and a VPS with docker-hosted Zrok endpoint

My zrok API endpoint is pointed to zrok subdomain of my VPS

In ubuntu server I started zrok agent service, following the official guide, and also I created my own user-service (zrokstartup.service):

[Unit]
Description=Zrok auto startup services, relies on zrok Agent
After=network-online.target
Wants=network-online.target

[Service]
Type=simple
User=kazekagyee
Environment=HOME=/home/kazekagyee
Environment=PATH=/usr/local/bin:/usr/bin:/bin
WorkingDirectory=/home/kazekagyee
ExecStart=/usr/local/services/zrok-start-services.sh
Restart=always
RestartSec=5

[Install]
WantedBy=multi-user.target

As you can see, it starts shell script zrok-start-services.sh:

#!/usr/bin/env bash

/usr/local/bin/zrok share reserved "someservice" &

wait

For a some time Zrok Agent behaves as expected, it starts shares (i have a number of them in zrok-start-services.sh) and maintaining access to them

And for some reason, at random moment of time Zrok Agent just drops all shares, until I access agent console or restart a whole machine it is running on

So I need some advice on troubleshooting this and accessing more detailed logs of Zrok Agent than zrok agent status provides, because I heavily rely on zrok as a proxy to access my homelab stuff, and also need it all to work after system start, not a big fan of staring at screen 5 seconds after waking up :laughing:

2

Hi @kazekagyee, welcome to the community and to zrok (and OpenZiti)!

I wouldn't be a fan of restarting the machine either. Can you capture the logs and inspect them to see if anything stands out that is causing the error? This sounds like a strange problem indeed, we'll need to see if there are any logs that can help figure out what the problem might be. Can you share the output of the agent when it's running and then when it goes 'bad'?

1 Like
3

There is a limitation in the v1 version of the zrok agent that when a share or access it is managing encounters an error, the share/access is removed from the agent… it has no notion of a share or access process in an “error” state, and no mechanism for retries.

We’ve soft-launched the new v2 version of zrok in production on zrok.io. The new v2 agent fully supports errored shares and handles retries.

Some of the semantics of how things work in v2 are different… specifically, reserved shares are handled in a completely new way.

There is a video available that demonstrates the new v2 features:

And there is a migration guide available for v2:

1 Like
4

Thanks for the reply!
After updating zrok-agent package to 1.1.10-1 it seems to work flawlessly (-ish, sometimes I get 502 error, but I thinks it’s my zrok’s node fault), and I got no problems with it so far

I’ll keep an eye on the agent’s behavior for a while and update this post if the issue happens again

P.S. I absolutely love this tool and your work for the community!

1 Like
5

Thanks for reply, I’ll definitely look forward to updating it!