I create VPS with API in Vultr Cloud. I have UserScript for a year and half. (UserScript is executed when VPS starting) That UserScript first installs docker
apt install --yes docker.io
then pulls image and runs it.
Last week apt install --yes docker.io
started failing.
Partial fix for the issue is:
ip link add name docker0 type bridge
ip addr add dev docker0 172.17.0.1/16
sleep 30
systemctl restart docker
If I create new VPS manually without UserScript and type apt install --yes docker.io
then I still get error.
I tried most recent Ubuntu 21.10 and there is same issue. I tried snap
instead of apt
and I get same issue.
But if I run Ubuntu 20.04 in AWS then I still can install docker with apt install --yes docker.io
Same true for Linode and DigitalOcean.
It seems no such problem with Fedora in Vultr.
I noticed some changes how Vultr assign IP addresses recenly.
So it seems they broke something with recent change and I contacted them but I got unexpectedly unhelpful response that Vultr is self managed service and they will not help with issues like docker.
Any ideas what may course the problem?
I do not post error logs as I already posted fix and logs are big.
Error initializing network controller: list bridge addresses failed: PredefinedLocalScopeDefaultNetworks List
May be some other Vultr clients that use docker in Ubuntu here?