When a container deployment fails as a result of a runtime error, the link creation stage should provide a clear error message instead of failed to Statfs "/proc/0/ns/net

[jbemmel]

For a Linux container I added the following cmd line:

cmd: 'sh -c "ip route replace default via 10.2.77.1"'

During initial startup, the subnet was unavailable so the command failed:

jvanbemmel@host:~$ docker logs clab-h1
ip: RTNETLINK answers: Network unreachable

However, that's not the error I get from clab. Instead:

ERRO[0000] failed to Statfs "/proc/0/ns/net": no such file or directory

[hellt]

this error means that this (or another) container existed before we reached the link creation phase.
Did it?

[jbemmel]

Yes, it exists briefly and then terminates

[jbemmel]

cmd: "false" may reproduce the issue

[hellt]

this is "expected", since containers should not exit normally
You should use exec to run commands inside the containers, and cmd is to change the daemon command that container runs

[jbemmel]

I made a mistake, but the error message could be clearer. Containerlab shouldn't even try to access an exited container