Hi, thanks for the report!
On Sun, Apr 23, 2017 at 12:31:02AM +0200, Baptiste Jonglez wrote: > I am using LXC on stretch, running a number of stretch containers. > > When running around 15 containers, systemd starts to fail, both on the host > and inside containers: > > # journalctl -f > Failed to get journal fd: Too many open files > # systemctl stop nginx > Failed to allocate directory watch: Too many open files > > The problem seems to be that systemd uses inotify quite a lot, and the limit > on inotify listeners is very low by default: > > # sysctl fs.inotify.max_user_instances > fs.inotify.max_user_instances = 128 > > Increasing this value indeed solves the problem. The issue was described > in this blog post: > > https://kdecherf.com/blog/2015/09/12/systemd-and-the-fd-exhaustion/ > > Attached is a patch for lxc that configures this sysctl setting to a higher > value. I don't know if this is the "Debian way" or if some other solution > is preferred. Your patch looks quite similar to what was added in LXD in Ubuntu recently: https://github.com/lxc/lxd-pkg-ubuntu/commit/333e2671a91861b949a2b990459656a3b065e5fb https://bugs.launchpad.net/juju/+bug/1602192 I think we also should add something like sysctl -q -p /etc/sysctl.d/lxc.conf || true to the configure target of postinst, to actually load the file directly, instead of waiting for a reboot? However, I am not 100% sure that packages should alter sysctl settings without telling the admin. Regards Evgeni
