lxd-to-incus fails on void linux #625
Labels
Incomplete
Waiting on more information from reporter
Comments
|
What version of Incus is that? I certainly remember implementing and testing void support in lxd-to-incus but maybe that change isn't in your version yet. |
0.5.1, the latest version in the void repos. I did see this PR but made this issue cause it seems the issue persists. If there is any logs or information you need, lmk. Thank you for the timely response. |
|
Would seem as though the void repo is out of date |
|
https://github.com/lxc/incus/actions/runs/8304297262/artifacts/1331261213 should get you the static binaries for the current main branch here which should include the latest lxd-to-incus |
The world needs more developers like you bredda. Cheers |
|
Looks like it is in for a PR. My apologize for not investigating this more |
|
No worries, glad it's working with the current version! |
even after deleteing the default zfs storage pool that was on incus, it then says: This is on incus 0.6 |
|
With the current version of the code, the only way this would happen is if Can you show:
|
|
|
Right so that's indeed not a clean Incus server, do:
And then run |
Thank you bredda. Just needed clarification if that was how I was supposed to proceed. Last issue I am encountering on void, after doing that when I try to start the container I get: |
|
Can you show |
|
|
There seems to be something going on with the Incus build on void, either because of the C library used or because of the kernel which is breaking pidfds. That's effectively out of scope for us as that's a distro-specific issue so something you may need to report to the void packager for Incus. That said, given your config above, I'd recommend doing:
Which should take you away from using |
if this is something leftover from a bad lxd-to-incus build can i just rm these and re-add them maybe? Luckily i only have to do the lxd to incus transition one time haha. |
|
Ah, that's interesting, I thought we did support wildcard listen address for NAT mode. If not, changing the |
|
When using LXD I was just making it so incomming on port 1337 would forward to port 9200 inside the container. My IP may change at times so thats why I was using 0.0.0.0 |
|
Okay, so yeah, you'd definitely benefit from I don't know much about void, but all our tests for setups like yours are passing fine so it's got to be something going on with void. Kernel is unlikely as that's not a kernel build option and your kernel is pretty recent, so something related to the C library would be my guest. Do you know if your system is using musl or glibc? |
|
glibc, yes. |
|
Im not sure if this helps: Kind of boned right now. All my containers have been converted with lxd-to-incus, just can't get them to start right now, so everything is halted. |
|
Very unfamiliar with this territory @stgraber , any other logs or debug information that might help? All my infrastructure is kind of stuck right now since it removed LXD already so I am at a stand still on this one :( |
|
Can you do: I've tested that here inside of a void container running on my Debian 12 system and that's working just fine, so if that's failing for you, then that would point towards a kernel issue. |
It looks like that ran no problem. side note: I still have my storage pool for the elastic container....so my data is ok i hope Thats so odd though. So what do you think, is there a better solution for the elasticcontainer? Any way I can maybe clone the container and add the port forwards to the new cloned container maybe? |
|
Can you try starting your container without those two devices, see if it starts up fine then or if it hits another problem? |
It ran without an error but the container is showing as STOPPED still
|
|
I suspect the container instantly dying on startup is actually the root cause of your problems as that would cause |
Weird, it's trying to look for a systemd mount...Void does not use systemd Cheers though...seems like we are getting somewhere on this....though this error message seems oddly familiar I am wondering, do you think its a cgroups issue with a difference in the host / container cgroup? |
|
Yeah, void not using systemd is likely to be the issue because that means that the required systemd cgroup wouldn't exist. Can you show:
|
|
There are different ways around this one but it depends on what void may already have set up. |
|
|
Ah, so hybrid v1 and v2, that's getting pretty unusual these days... |
CHEERS. Holy crap that was a nightmare lol. @stgraber I will say it again, you are one of the most helpful & reactive developers I know of. Thank you bredda. |
|
in case it helps, void specific stuff in usually in readme.void
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
incus infoandlxc infoboth working, didn't initialize incus though (as per the documentation)The text was updated successfully, but these errors were encountered: