Where is Systemd Boot time messages?
FedoraForum.org - Fedora Support Forums and Community
Results 1 to 14 of 14
  1. #1
    Join Date
    Dec 2011
    Location
    Kerala,India
    Posts
    58
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Question Where is Systemd Boot time messages?

    I`m using fed17. It take unusually too much time for gnome login screen to show up. It seems like the system is hanging up at
    Code:
     [OK] Started Recreate Volatile Files and Directories
    . It fails everytime. After some timeout has reached (bcoz i see no hard disk activity during this time), things load up as usual

    Also, there is some line like
    Code:
    [DEPEND] <something>
    This is printed immediately after the "Failed to recreate volatile files and directories" error message. But as the screen is scrolling very fast at this moment, i cant read the "DEPEND" part. It must contain a detailed information for the failure.
    The output of
    Code:
    systemctl status 'tmpfiles-setup.service'
    is:

    Code:
    tmpfiles-setup.service
    	  Loaded: error (Reason: No such file or directory)
    	  Active: inactive (dead)
    So where do I find the systemd log files that contain the boot time messages we see when hitting DEL key at plymouth splash screen??

    Further, this anomaly occurred after installing another Linux distro in such a way that , the existing installation (fed17)`s partition number changed to sda5 from some other sdax ( i mean x != 5). yes it really did. I keep trying different distros and my partitions were really messed up somehow


    --
    Thanks in advance,
    Jay Aurabind

  2. #2
    Join Date
    May 2011
    Location
    Confoederatio Helvetica (Swissh)
    Age
    40
    Posts
    4,370
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    dmesg ?
    systemd-analyze blame ?

    Code:
    for tmp in ERROR DEPEND;do echo "$tmp:";dmesg|grep $tmp;done;systemd-analyze blame;systemd-analyze
    Last edited by sea; 3rd November 2012 at 06:09 PM.
    * Yet Another Simple Script Installer (YASSI): https://forums.fedoraforum.org/showthread.php?323099
    * Shell Wrapper and Runtime Modifier (SWARM): <not yet usable>

    * EFI Cheatsheet :: http://forums.fedoraforum.org/showthread.php?t=298546
    * Windows 8+ & Fedora 20+ Dualboot :: http://forums.fedoraforum.org/showthread.php?t=298161

  3. #3
    Join Date
    Aug 2009
    Location
    Waldorf, Maryland
    Posts
    7,345
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    If they don't show up in /var/log/messages or from dmesg, I expect systemd lost them.

  4. #4
    Join Date
    Dec 2011
    Location
    Kerala,India
    Posts
    58
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    Thank you sea and jpollard for the reply.

    the output of
    Code:
    for tmp in ERROR DEPEND;do echo "$tmp:";dmesg|grep $tmp;done;systemd-analyze blame;systemd-analyze
    is:

    Code:
    ERROR:
    [    0.229064]  pci0000:00: ACPI _OSC request failed (AE_ERROR), returned control mask: 0x1d
    DEPEND:
     11641ms udev-settle.service
      4790ms fedora-loadmodules.service
      2765ms systemd-binfmt.service
      2501ms fedora-storage-init-late.service
      2127ms systemd-vconsole-setup.service
      1917ms fedora-storage-init.service
      1823ms fedora-readonly.service
      1617ms systemd-remount-fs.service
      1606ms media.mount
      1575ms proc-sys-fs-binfmt_misc.mount
      1563ms dev-hugepages.mount
      1532ms dev-mqueue.mount
      1517ms colord-sane.service
      1508ms sys-kernel-debug.mount
      1357ms sys-kernel-config.mount
      1081ms akmods.service
       845ms udev-trigger.service
       781ms dkms_autoinstaller.service
       738ms udev.service
       661ms NetworkManager.service
       655ms colord.service
       635ms lvm2-monitor.service
       626ms systemd-sysctl.service
       580ms mdmonitor-takeover.service
       386ms systemd-logind.service
       362ms avahi-daemon.service
       295ms dbus.service
       285ms livesys.service
       282ms systemd-readahead-collect.service
       276ms iptables.service
       266ms ip6tables.service
       265ms systemd-readahead-replay.service
       260ms systemd-user-sessions.service
       244ms auditd.service
       222ms irqbalance.service
       220ms mcelog.service
       215ms systemd-tmpfiles-setup.service
       210ms ksm.service
       188ms fedora-wait-storage.service
       165ms ksmtuned.service
       164ms abrt-vmcore.service
       163ms abrt-ccpp.service
        99ms jexec.service
        52ms udisks2.service
        47ms bluetooth.service
        43ms spice-vdagentd.service
        32ms livesys-late.service
        26ms accounts-daemon.service
        16ms upower.service
        13ms xinetd.service
         4ms sys-fs-fuse-connections.mount
         4ms rtkit-daemon.service
    Startup finished in 1765ms (kernel) + 4747ms (initramfs) + 94462ms (userspace) = 100975ms
    Seems like the real message I want is not showing up anywhere. I`m not convinced with all those "depends" being showed. It appeared in yellow color at the boot time so I`m sure that there is ONLY ONE "depend". So what depends are these ?

    Moreover, the real error i got are missing, just like jpollard said. I wonder computers have amnesia :P

    Do plymouth have any special log? I cant find anything in dmesg

  5. #5
    Join Date
    Sep 2005
    Location
    Redneck Riviera
    Posts
    408
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    I believe what you're looking for is /var/log/boot.log

    Also, the "Recreate Volatile Files and Directories" message looks like it's from /lib/systemd/system/systemd-tmpfiles-setup.service
    Last edited by cazo; 4th November 2012 at 03:50 AM. Reason: systemd-tmpfiles-setup.service

  6. #6
    Join Date
    May 2011
    Location
    Confoederatio Helvetica (Swissh)
    Age
    40
    Posts
    4,370
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    Nah those lines are the output of "systemd-analyze blame".
    I wanted to see the actual output where you stated "<something>", but it seems its on another line.

    Try this instead:
    Code:
    for tmp in ERROR DEPEND;do echo "$tmp:";dmesg|grep $tmp -A3;done
    * Yet Another Simple Script Installer (YASSI): https://forums.fedoraforum.org/showthread.php?323099
    * Shell Wrapper and Runtime Modifier (SWARM): <not yet usable>

    * EFI Cheatsheet :: http://forums.fedoraforum.org/showthread.php?t=298546
    * Windows 8+ & Fedora 20+ Dualboot :: http://forums.fedoraforum.org/showthread.php?t=298161

  7. #7
    Join Date
    Dec 2011
    Location
    Kerala,India
    Posts
    58
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    Hurray! found it! Thank you Cazo!!!

    These are the interesting lines in it:

    Code:
    [ TIME ] Timed out waiting for device dev-disk-by\x2duuid-44033812\x2dd8fb\x2d4537\x2da724\x2d902e3cd54b77.device.
    [DEPEND] Dependency failed for /dev/disk/by-uuid/44033812-d8fb-4537-a724-902e3cd54b77.
    Look like a linux install altered the UUID of my swap partition. Fixed it in /etc/fstab and its all working great now.

    By the way, Sea, I think "Driver Messages" doesnt fit into Boot messages. Maybe thats why it is not showing up in dmesg.

    Thank you very much experts! The problem is now solved!

  8. #8
    Join Date
    Aug 2009
    Location
    Waldorf, Maryland
    Posts
    7,345
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    Quote Originally Posted by aurabindo
    Thank you sea and jpollard for the reply.
    ...
    Seems like the real message I want is not showing up anywhere. I`m not convinced with all those "depends" being showed. It appeared in yellow color at the boot time so I`m sure that there is ONLY ONE "depend". So what depends are these ?

    Moreover, the real error i got are missing, just like jpollard said. I wonder computers have amnesia :P
    Not amnesia - bad design. It used to be that one of the first things initialized was root mount, then syslog. Any messages sent to the console were in a memory buffer and syslog would dump these. Severe errors would sometimes cause the buffer to overflow (loosing the first messages) but that was rather rare. Systemd starts everything (mostly) in parallel, which can overflow the buffer MUCH faster (now there is systemd+dbus+otherstuff being done at once) and the message logs get lost. Could also be a side effect of systemd not responding fast enough to catch some, though I don't think that happens after syslog is running.

    Do plymouth have any special log? I cant find anything in dmesg
    No - at that time the only log buffer I know of is still the kernel log buffer attached to the console device. Note: plymouth is running BEFORE systemd is started - so the only thing carried across is the kernel log buffer/console output. When that is filled it starts recycling, and can loose data.

    My dmesg output never seems to have anything from the first two tenths of a second log, so anything before (memory problems or whatever did happen) is lost. Granted, I do have a fairly large system with lots of USB controllers (11) 8 cores, 8GB memory, plus 8 SAS (unused), 5 SATA (two used), PATA (old disk+DVD) and even a floppy, so there are a fair number of messages.

    Part of the problem is that the initrd generates the first messages - and this can/will fill a buffer quite quickly, and there is no place to record this outside memory (unless you are configured to use a serial console...) until after root gets mounted read/write AND have syslog running.

    Unfortunately, that now requires having systemd+dbus operational (and any messages that generates). Now the systemd-kmsg-syslogd.service has to be started (it "bridges" the kernel log, so should get it copied out for recording..) but it has no dependencies, and syslog has to wait for it. Note - syslog.service is a socket... so the actual syslog process doesn't start until AFTER messages are queued. And a UDP socket has only a limited number of entries before new ones get discarded... and syslog is started along with the rest of the thundering herd of startup processes.

    The only thing that waits for syslog happens to be loader.service ... which on my system doesn't exist. To me, that means nothing can be relied upon for logging messages properly.

    So anything that overflows the buffers (wherever they may be) in between appears to have the potential to be lost.
    Last edited by jpollard; 4th November 2012 at 01:33 PM. Reason: bad day for syntax...

  9. #9
    Join Date
    Aug 2009
    Posts
    8,586
    Mentioned
    1 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    You can try looking in the journal and see if the messages are in there instead of syslog. systemd puts a lot of messages into it's journal file now.

    run journalctl to display the journal messages. It has quite a few options if you wish to look at them. man journalctl

  10. #10
    Join Date
    Dec 2011
    Location
    Kerala,India
    Posts
    58
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    Hail jpollard! Thats a lot of information. Thank you for your patience to type it in!

    And it seems journalctl is the official log. Thank you DBelton

  11. #11
    Join Date
    Aug 2009
    Location
    Waldorf, Maryland
    Posts
    7,345
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    It's too bad systemd broke the logging...

    It used to be possible to have non-local logs for audit reasons.

    It seems that systemd cannot do this as it now requires a local log...

  12. #12
    Join Date
    Jul 2004
    Location
    Halifax
    Posts
    110
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Thumbs down Re: Where is Systemd Boot time messages?

    systemd has broken a lot of things $*!. I miss LSB. I did a yum search of journalctl and it did not show up. F16 here. I need to figure out what is stopping my booting process, and since everything starts in parallel, it's impossible to tell what's stopping it. Really, really, pissed off for fedora moving to this, for a measly 20sec boot time

  13. #13
    Join Date
    Jun 2004
    Location
    Maryland, US
    Posts
    8,077
    Mentioned
    16 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    A while back, there used to be a tool called "plymouth-log-viewer" but it doesn't exist any more in the plymouth tools

  14. #14
    Join Date
    Aug 2009
    Location
    Waldorf, Maryland
    Posts
    7,345
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    Re: Where is Systemd Boot time messages?

    Even the 20sec boot time is illusory - It is entirely possible to get 20 second boots with SysVinit. Just don't put so many scripts in the startup sequence (even disabled, they get scanned just to be skipped, better if they weren't there in the first place).

Similar Threads

  1. Analyze the startup time of your OS with systemd
    By salemeni in forum Guides & Solutions (Not For Questions)
    Replies: 3
    Last Post: 17th August 2011, 06:28 PM
  2. [SOLVED]
    systemd-fsck superblock last write time on each boot
    By lmcogs in forum F15 Development
    Replies: 10
    Last Post: 17th April 2011, 01:17 PM
  3. FC 13 has long boot time, Boot Messages Posted
    By teknodude in forum Using Fedora
    Replies: 3
    Last Post: 26th August 2010, 03:34 AM
  4. system time in /var/log/messages
    By amitbern in forum Using Fedora
    Replies: 4
    Last Post: 17th March 2009, 07:32 AM

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •