Warning in the container log

Hi,
I have LXD 3.0.3 which comes from Ubuntu server 18.04.1.

I did following:

lxc launch images:centos/6/i386 test-centos6
lxc info --show-log test-centos6

and got following warning:

Log:

lxc test-centos6 20190215102455.365 WARN conf - conf.c:lxc_setup_devpts:1616 - Invalid argument - Failed to unmount old devpts instance

if I stop the container and do again : lxc info --show-log test-centos6
will get a little more info in the log:

Log:

lxc test-centos6 20190215102455.365 WARN conf - conf.c:lxc_setup_devpts:1616 - Invalid argument - Failed to unmount old devpts instance
lxc test-centos6 20190215111204.382 WARN network - network.c:lxc_delete_network_priv:2589 - Operation not permitted - Failed to remove interface “eth0” with index 92

any idea why? Thanks

Both of those warnings are fine to ignore when driven by LXD.

@brauner either of those warrant a fix somewhere in liblxc?

We could dumb them down from warnings to traces. We might even remove the log entry for the failed devpts umount.

Oh, the errno doesn’t make sense at least EPERM is wrong that’s just because we return -1 from netlink functions. I’ll fix that.

I re-installed the lxd with snap to get the latest version, 3.10, and did same test, it shows more error:
Log:

lxc test-centos6 20190215235340.942 WARN conf - conf.c:lxc_map_ids:2970 - newuidmap binary is missing
lxc test-centos6 20190215235340.942 WARN conf - conf.c:lxc_map_ids:2976 - newgidmap binary is missing
lxc test-centos6 20190215235340.961 WARN conf - conf.c:lxc_map_ids:2970 - newuidmap binary is missing
lxc test-centos6 20190215235340.961 WARN conf - conf.c:lxc_map_ids:2976 - newgidmap binary is missing
lxc test-centos6 20190215235341.124 WARN conf - conf.c:lxc_setup_devpts:1641 - Invalid argument - Failed to unmount old devpts instance

@brauner newuidmap/newgidmap warning should probably go away when we’re running as root (and be a fatal error when not).

Hi,
in the LXD 3.01/3.03:
lxc launch ubuntu:18.04
log will have no error, now a brand new 3.10 , it shows error as well:
Log:

lxc shining-feline 20190216000820.452 WARN conf - conf.c:lxc_map_ids:2970 - newuidmap binary is missing
lxc shining-feline 20190216000820.452 WARN conf - conf.c:lxc_map_ids:2976 - newgidmap binary is missing
lxc shining-feline 20190216000820.493 WARN conf - conf.c:lxc_map_ids:2970 - newuidmap binary is missing
lxc shining-feline 20190216000820.493 WARN conf - conf.c:lxc_map_ids:2976 - newgidmap binary is missing
lxc shining-feline 20190216000820.596 WARN conf - conf.c:lxc_setup_devpts:1641 - Invalid argument - Failed to unmount old devpts instance

Yesterday I had to run updates on my host system and got an upgrade to LXD 3.21.

Now I’m having exactly the same problem reported by the OP, and the container won’t start at all. I typically run lxc commands as myself on this system, am i reading that I need to start the container as root for this problem to go away? Also just to confirm: this container has been running in production for months, so it was definitely the upgrade that caused this (unfortunately not sure what versions of glibc, LXD, etc. I upgraded from because I had to clear the package cache).

Oops; so sorry. I just realized I necro-bumped this ticket (forgot that it’s 2020, not 2019 – the month fooled me). Will top post.