zfs + lxc + docker problems (i think)

[Kieeps]

Im trying to setup nextcloud + collabora withing docker in a LXC container but i keep running in to a problem that might be a due to the lxc + zfs combo.

I figured that maby someone else on these forums know how to fix this?

if i install collabora in a VM instead it works fine, if i run it localy on my laptop (runs arch btw) it also works fine, it's only whe i try to run it in the dedicated docker lxc :/

i pasted the output here https://controlc.com/50c3b993

but basically i get this:

[36mcollabora |[0m frk-00033-00033 2021-12-09 07:57:49.543456 +0000 [ coolforkit ] ERR Capability cap_mknod is not set for the coolforkit program.| kit/ForKit.cpp:226
[36mcollabora |[0m frk-00033-00033 2021-12-09 07:57:49.543460 +0000 [ coolforkit ] ERR Capability cap_fowner is not set for the coolforkit program.| kit/ForKit.cpp:226
[36mcollabora |[0m frk-00033-00033 2021-12-09 07:57:49.543464 +0000 [ coolforkit ] ERR Capability cap_chown is not set for the coolforkit program.| kit/ForKit.cpp:226

[36mcollabora |[0m wsd-00001-00032 2021-12-09 07:57:54.546419 +0000 [ prisoner_poll ] TRC Poll completed with 0 live polls max (5000000us)(timedout)| net/Socket.cpp:377
[36mcollabora |[0m wsd-00001-00032 2021-12-09 07:57:54.546461 +0000 [ prisoner_poll ] DBG Starting handling poll results of prisoner_poll at index 0 (of 1)| net/Socket.cpp:453
[36mcollabora |[0m wsd-00001-00032 2021-12-09 07:57:54.546471 +0000 [ prisoner_poll ] TRC ppoll start, timeoutMicroS: 5000000 size 1| net/Socket.cpp:358
[36mcollabora |[0m wsd-00001-00032 2021-12-09 07:57:59.550866 +0000 [ prisoner_poll ] TRC Poll completed with 0 live polls max (5000000us)(timedout)| net/Socket.cpp:377
[36mcollabora |[0m wsd-00001-00032 2021-12-09 07:57:59.550908 +0000 [ prisoner_poll ] DBG Starting handling poll results of prisoner_poll at index 0 (of 1)| net/Socket.cpp:453
[36mcollabora |[0m wsd-00001-00032 2021-12-09 07:57:59.550915 +0000 [ prisoner_poll ] TRC ppoll start, timeoutMicroS: 5000000 size 1| net/Socket.cpp:358

root@kloudr:~# cat /etc/pve/lxc/101.conf

arch: amd64
cores: 8
features: fuse=1,nesting=1
hostname: docker
memory: 8192
mp1: /Tank,mp=/mnt/Tank
mp2: /rpool/Docker,mp=/mnt/Docker
net0: name=eth0,bridge=vmbr1,firewall=1,gw=192.168.2.1,hwaddr=42:61:BF:E36:EE,ip=192.168.2.100/24,type=veth
ostype: debian
rootfs: local-zfs:subvol-101-disk-0,size=32G
swap: 8192

PVE Version 7.1-6



please let me know if there are any info i'w missed.

 

[LnxBil]

I figured that maby someone else on these forums know how to fix this?

You mean besides not using Docker and LXC together? No.

This has been discussed over and over again by others and my me: Do not use LXC + Docker unless you know what you're doing. Just use a VM, install a Docker host os of you choice and just go with that.

I use also ZFS+LXC+Docker but i regularely run into problems, so I can hear you.

 

[Kieeps]

Really? I'w never really run in to a problem with lxc + docker before, i managed to solve this specific problem by running it on LVM instead of zfs.
Apparently you can run it on zfs if you configure docker to run with the device mapper driver, but when i tried it i got a warning that device mapper was deprecated so that didn't feal like a valid solution imo.

I have though about moving docker to a VM, but i like the smaller footprint of a LXC