[Feature request] Validate that the k3s-agent started with systemd about k3sup HOT 13 CLOSED

Thanks for the issue.

I'm pretty sure that everywhere I tested k3sup, I used Ubuntu 18.04.3 LTS. I'm surprised it isn't working... What cloud did you use?

As for checking if the agent started. Do you know any bash command that could do that?

from k3sup.

I’m an using freshly installed VMs on my local KVM machines.

For machines with systemd...

systemctl status application.service

from k3sup.

You could add a function validateAgent in join.go So they you could validation code.

I’m sure there are other things worth validating.

I don’t know Go we’ll enough or I would create a branch/PR.

from k3sup.

That tends to be blocking? Does it need a human to read it, or can it be run over SSH with an exit code being parsed?

Feel free to PoC something.

from k3sup.

If you ware wanting something more pure Bash ...

 ps -ef | grep "k3s agent" | grep -v grep > /dev/null

... has an exit code of 0 on a match, and 1 on no match.

from k3sup.

It seems a bit brittle. Is there an FD created maybe when systemd starts up a process?

Maybe you could ask upstream for a shell script to by provided by k3s, just like they provide an uninstaller.sh

from k3sup.

I got the same thing on Amazon Linux 2 following https://rancher.com/blog/2019/k3s-kubeconfig-in-seconds/.

After the exec error:

Sep 02 18:35:38 ip-172-31-0-204.ec2.internal systemd[1]: [/etc/systemd/system/k3s-agent.service:7] Failed to parse service type, ignoring: exec

This error is logged:

Sep 02 18:36:21 ip-172-31-0-204.ec2.internal k3s[3958]: time="2019-09-02T18:36:21.489789359Z" level=error msg="failed to get CA certs at https://127.0.0.1:42269/cacerts: Get https://127.0.0.1:42269/cacerts: read tcp 127.0.0.1:47350->127.0.0.1:42269: read: connection reset by peer"

https://127.0.0.1:42269 hangs.

The second one is covered by
k3s-io/k3s#484

In particular for the second one, make sure there is network connectivity between the agent and server hosts.

The first one seems to be a k3sup thing. I reinstalled using the regular k3s install.sh and it worked ok.

from k3sup.

There should be no error @sourcedelica if you're following the tutorial. Please can you raise a separate issue about the tutorial, then I'll ping the author to see if he can give you some technical support.

Alex

from k3sup.

@dhoard does the ssh / exec return a non-zero exit code in your example, or can it only be determined through manual checking?

from k3sup.

@alexellis not sure... but since the systemd start is asynchronous, I suspect you can’t use exit code.

from k3sup.

@sourcedelica I just tried again for setting up k3s using k3sup on Amazon Linux 2 using the blog and it works without giving any error. I did not do any additional step apart from what is mentioned in the blog. Please make sure you have set up the ports, running as root etc. It is very straightforward so not sure why its giving error for you. Feel free to create another issue for the same as mentioned by @alexellis earlier so that we can have a detailed discussion on it.

from k3sup.

Sounds good, thanks

from k3sup.

Not all Linux distros are the same. I urge that this ticket be reopened or forked... I ran the install and join commands against VMware's PhotonOS v3 and had the same two errors.

• systemd type:exec error
• CA File missing and certs

If this is something that the user is supposed to do then a link to the docs would be helpful because not even the k3s docs mention these issues.

from k3sup.