Fedora Account System
Red Hat Associate
Red Hat Customer
+++ This bug was initially created as a clone of Bug #833156 +++ Description of problem: error messages displayed during a reboot are 'lost' at shutdown They should persist by being saved into the log files, so they may be studied and resolved error message showing , in red: [ OK ] Stopped Set time via NTP. Starting Show Plymouth Reboot Screen... Stopping Permit User Sessions... [FAILED] Failed to start Show Plymouth Reboot Screen. See 'systemctl status plymouth-reboot.service' for details. [ OK ] Stopped System Logging Service Version-Release number of selected component (if applicable): systemd-44-12.fc17.armv5tel How reproducible: VFAD 2 test day, vexpress image, updated with yum and them rebooted Steps to Reproduce: as above Actual results: [ OK ] Stopped Set time via NTP. Starting Show Plymouth Reboot Screen... Stopping Permit User Sessions... [FAILED] Failed to start Show Plymouth Reboot Screen. See 'systemctl status plymouth-reboot.service' for details. [ OK ] Stopped System Logging Service Expected results: persistence of the error message, through a reboot, via placing errors into /var/log/messages so long as the syslogd is still running (which should be quite late in a shutdown process, to permit catching such) Additional info: This bug cloned from its parent, and implies a need for an RFE to systemd, to persist error messages across reboots, probably into: var/log/messages
The journal stays around until very late and will show you this data on F18 (journalctl). If you want that rsyslog stays longer around to collect this data, please file a separate bug so that they mark their service file as special, so that it can stay around.
reading comment 1 seemingly, an addition for rsyslog in its systemd configuration file is needed to persist running until the RW filesystem is umounted
Please try to switch your journald to persistent storage, this should help to fix your problem (Storage=persistent in /etc/systemd/journald.conf).
@Jiří Vymazal thank you for the comment [herrold@centos-7 ~]$ grep -v ^# /etc/systemd/journald.conf [Journal] [herrold@centos-7 ~]$ grep -i "storage" /etc/systemd/journald.conf #Storage=auto [herrold@centos-7 ~]$ rpm -qf /etc/systemd/journald.conf systemd-219-42.el7_4.4.x86_64 [herrold@centos-7 ~]$ I dunno that a fix for my particular configuration, or a smarter 'auto' setting in systemd is the issue in play testing
reading the man page, it seems also a file needs to be touched, in addition to moving from 'auto' to 'persistent': [root@centos-7 ~]# man 5 journald.conf [root@centos-7 ~]# ls -al /var/log/journal ls: cannot access /var/log/journal: No such file or directory [root@centos-7 ~]# touch /var/log/journal [root@centos-7 ~]# restorecon -Rv /var/log/journal one concern: Will this be picked up by logrotate, or grow without limit? still testing
I am going to close this as I believe that in latest fedora+rsyslog, and using persistent storage this issue is fixed, in case of problems please reopen/file new report.