Activity log for bug #1803086

Date Who What changed Old value New value Message
2018-11-13 08:42:19 Chris Malton bug added bug
2018-11-13 08:42:19 Chris Malton attachment added journalctl -u systemd-networkd https://bugs.launchpad.net/bugs/1803086/+attachment/5212217/+files/systemd-networkd.log
2018-11-13 08:43:17 Chris Malton description *Somewhere* it looks like netplan reloads systemd-networkd weekly, leading to IP addresses added by Keepalived (and other similar programs) "evaporating" from interfaces once a week. I would report this as a bug in systemd-networkd, but Debian does not have this problem and Ubuntu does, and the difference is netplan + systemd, vs /e/n/i + systemd so I am currently believing this is a netplan bug (I can't find any suggestion anywhere on my system that systemd-networkd is reloading itself). There are a few reports of this online (mainly on the keepalived mailing list and bug tracker - Notably https://github.com/acassen/keepalived/issues/836) but not many. Example log from systemd-networkd is attached - in my case both eth0 and eth1 have IPs managed by keepalived. I don't get these log messages on Debian, so believe it's something Ubuntu specific and likely to be netplan related. If not, apologies, and please let me know in which project I should open a bug report. Unfortunately, of the back of this, I'll be switching all our load balancers to Debian rather than Ubuntu, largely down to the fact that Ubuntu's standard repos don't ship a version of Keepalived that can cope with this and recover from it. *Somewhere* it looks like netplan reloads systemd-networkd weekly, leading to IP addresses added by Keepalived (and other similar programs) "evaporating" from interfaces once a week. I would report this as a bug in systemd-networkd, but Debian does not have this problem and Ubuntu does, and the difference is netplan + systemd, vs /e/n/i + systemd so I am currently believing this is a netplan bug (I can't find any suggestion anywhere on my system that systemd-networkd is triggering its own reload). There are a few reports of this online (mainly on the keepalived mailing list and bug tracker - Notably https://github.com/acassen/keepalived/issues/836) but not many. Example log from systemd-networkd is attached - in my case both eth0 and eth1 have IPs managed by keepalived. I don't get these log messages on Debian, so believe it's something Ubuntu specific and likely to be netplan related. If not, apologies, and please let me know in which project I should open a bug report. Unfortunately, of the back of this, I'll be switching all our load balancers to Debian rather than Ubuntu, largely down to the fact that Ubuntu's standard repos don't ship a version of Keepalived that can cope with this and recover from it.