diff options
| -rw-r--r-- | core/modules/dhcpc-busybox/data/etc/udev/rules.d/99-openslx-net.rules | 9 |
1 files changed, 8 insertions, 1 deletions
diff --git a/core/modules/dhcpc-busybox/data/etc/udev/rules.d/99-openslx-net.rules b/core/modules/dhcpc-busybox/data/etc/udev/rules.d/99-openslx-net.rules index 17c0068c..8aa11f0d 100644 --- a/core/modules/dhcpc-busybox/data/etc/udev/rules.d/99-openslx-net.rules +++ b/core/modules/dhcpc-busybox/data/etc/udev/rules.d/99-openslx-net.rules @@ -1 +1,8 @@ -ACTION=="add", SUBSYSTEM=="net", KERNEL=="br0|br-nic-*", TAG+="systemd", ENV{SYSTEMD_ALIAS}="/sys/subsystem/net/devices/$name", ENV{SYSTEMD_WANTS}="network-interface@$name.service" +# This doesn't reliably trigger. It's a rare race condition, on some machines it always works, on others it's +# maybe a 1:20 chance it doesn't work. Apparently udev tries to remember whether a device was already handled +# by systemd and ignores further add events. So if you think this problem doesn't exist anymore, make sure +# you're not just lucky and not hitting this issue, and also make sure you understand what the initial problem +# was. I assume it's something to with switching from initrd to final root, but this isn't worth the effort tbh. +#ACTION=="add", SUBSYSTEM=="net", KERNEL=="br0|br-nic-*", TAG+="systemd", ENV{SYSTEMD_ALIAS}="/sys/subsystem/net/devices/$name", ENV{SYSTEMD_WANTS}="network-interface@$name.service" +# So use a more crude approach that triggers every time (consecutive invocations shouldn't do anything) +ACTION=="add", SUBSYSTEM=="net", KERNEL=="br0|br-nic-*", RUN+="/usr/bin/systemctl --no-block start network-interface@$name.service" |