On Fri, 2017-11-17 at 17:10 +0000, Mark Stosberg wrote: > As noted in the linked bug report, there are some logs which are > uniquely stored in the systemd journal, the journal is not > persistent-by-default, and data loss is occurring as a default > behavior.
AFAICT rsyslogd wasn't special-cased in the upstart days, meaning it would get killed with other upstart services in semi-random order, so it never was able to log the very last messages. It would get killed after most services that were defined by a SysV init script though. I'm not sure how much of a regression there really is; it would depend on how early rsyslogd gets killed now... If journald uses some specific configuration/dependency to always be stopped as late as possible, then maybe that can be applied to the rsyslogd unit too, so that they get killed virtually together (or even rsyslogd last, if that's possible). That could solve your issue in current releases even? -- Jan Claeys -- ubuntu-devel mailing list ubuntu-devel@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel