Bugzilla – Bug 642289
System does not boot correctly with sysvinit-tools-2.88-15.1.x86_64.rpm or newer (blogd issue)
Last modified: 2011-10-11 09:49:13 UTC
User-Agent: Mozilla/5.0 (compatible; Konqueror/4.5; Linux) KHTML/4.5.1 (like Gecko) SUSE My system is configured to offer a graphical login screen (kdm) on startup. After updating to a recent Factory version of Opensuse the boot process does not continue until reaching graphical login. Instead a login prompt is offered. The issue does not end there: The system also failed to load the German keyboard mapping (can be restored manually). When starting KDE manually there are error messages not shown hitherto (e.g. there is a message that sound devices could not be found). After reinstalling some older packages I found that the package sysvinit-tools-2.88-15.1.x86_64.rpm is the culprit: When I downgrade it to version sysvinit-tools-2.88-14.1.x86_64.rpm everything works normally. I tried to narrow the possible origin of the problems down. To that end, I extracted the binaries from the newer sysvinit-tools and installed them one by one. I found that having /sbin/blogger and /sbin/blogd from the newer sysvinit-tools suffices to reproduce the problem. All other binaries can remain from the older version. I also went the reverse way. I reinstalled sysvinit-tools-2.88-15.1.x86_64.rpm (i.e. the newer version). I found that for the problem to go away it is enough to replace just /sbin/blogd and /sbin/blogger with the older version. The problem does also happen with sysvinit-tools-2.88-16.1.x86_64.rpm. From http://lists.opensuse.org/opensuse-commit/2010-08/msg00567.html I see that indeed there have been changes made to /sbin/blogd and /sbin/blogger. Could these be the origin of the problems? Reproducible: Always Steps to Reproduce: 1. Upgrade the system to a recent Factory revision (e.g. 30 August 2010, or 28 September)
This looks like a duplicate *** This bug has been marked as a duplicate of bug 642334 ***
This is not yet resolved. With sysvinit-tools-2.88-17.1.x86_64.rpm the system does still not start up completely. For correct startup, I can do the following: 1. Use /sbin/blogd from sysvinit-tools-2.88-14.1.x86_64.rpm, or 2. Set the sysconfig variable "RUN_PARALLEL" to "no".
I updated from opensuse 11.3 to 11.4 and wondered why the system did not init the services of runlevel 5 or 3. Manually running init 3 and then init 5 correctly started X / kdm etc. Now I set the sysconfig variable "RUN_PARALLEL" to "no", as mentioned above. If I figured out correctly, this is rather a workaround. Also regarding the duplicate bug, what are the actual / official steps to fix this ?
Use *latest* sysvinit-tools: last three changelog entries are - Make blogd work together with kernel from 11.4 even on a serial system console (bnc#672450) - Make option -k for killproc utilities work for normal users even if the exe link of an own process remains to root (bnc#664941) - New killproc version 2.19: bug fix update
Created attachment 420341 [details] screenshot from failed boot to runlevel 5 and failed init 0
At first this seemed to solve this issue, but after some time / multiple boot processes I encountered this problem again. I have installed > rpm -qv sysvinit-tools sysvinit-tools-2.88+-148.2.x86_64 with the last changlog entries being > rpm -q --changelog sysvinit-tools | head * Di Mär 15 2011 werner@suse.de - Update to current SVN version of sysvinit 2.88dsf: * Fix counting message lines in wall. Patch from Petr Lautrbach. * Fix bad printf conversion specifier in wall. Patch from Sébastien Luttringer. * Add patches from Openwall project........... and the ones you mentioned.
see also https://bugzilla.novell.com/show_bug.cgi?id=680332
Hmm... looks like my bug https://bugzilla.novell.com/show_bug.cgi?id=680332 is pretty the same. BUT this bug is marked as duplicate of https://bugzilla.novell.com/show_bug.cgi?id=642334 and #642334 DOES NOT LOOK LIKE MY PROBLEM!!! BEFORE YOU ANYONE SETS #680332 AS A DUPLICATE OF THIS BUG, MARK THIS AS >>>NOT RESOLVED<<< !!! Thanks :-) Just want to make sure, my bug doesn't gets lost :-)
This makes no sence to me: Why should blogd reading from fd 0 aka the current pseudo-terminal master of the system console terminal slave? And where comes the other errors from? I do not see such error on my own 11.4 at home. Question: What shell are you using for /bin/bash and /bin/sh?
Next question: What does stty size report on a virtual console and what does stty size < /dev/console report? What commands are included in /etc/init.d/before.local?
Next question: which frame buffer driver is used for the virtual consoles?
> Question: What shell are you using for /bin/bash and /bin/sh? I don't know which information you mean exactly. In /etc/passwd, bash is set (or false/nologin) and versions are: bash --version GNU bash, Version 4.1.10(1)-release (x86_64-suse-linux-gnu) Copyright (C) 2009 Free Software Foundation, Inc................. sh --version GNU bash, Version 4.1.10(1)-release (x86_64-suse-linux-gnu) ....... > And where comes the other errors from? I do not see such error on my own 11.4 at home. I'dont know. There did not occur problems with rc.status on successful boot. At the moment I use RUN_PARALLEL=false and did not encounter problems. I do not know the source of the mdraid message on shutdown (I have a raid 1). On tty1: >stty size 60 156 >stty size < /dev/console 60 156 On tty2 output is "64 160". > What commands are included in /etc/init.d/before.local? None, every line is commented. > Which frame buffer driver is used for the virtual consoles? dmesg | grep fb [ 0.000000] Command line: root=/dev/disk/by-id/md-uuid-c368d599:f686e139:c1c8f2ba:7d4de9a9 resume=/dev/disk/by-id/md-uuid-ea4defb0:d47b988c:c1c8f2ba:7d4de9a9 splash=silent vga=0x31a [ 0.000000] ACPI: RSDP 00000000000fb9e0 00024 (v02 ACPIAM) [ 0.000000] Kernel command line: root=/dev/disk/by-id/md-uuid-c368d599:f686e139:c1c8f2ba:7d4de9a9 resume=/dev/disk/by-id/md-uuid-ea4defb0:d47b988c:c1c8f2ba:7d4de9a9 splash=silent vga=0x31a [ 0.154481] pci ...... <cut> [ 0.344571] vesafb: framebuffer at 0xd0000000, mapped to 0xffffc90011100000, using 5120k, total 262144k [ 0.344577] vesafb: mode is 1280x1024x16, linelength=2560, pages=1 [ 0.344579] vesafb: scrolling: redraw [ 0.344582] vesafb: Truecolor: size=0:5:6:5, shift=0:11:5:0 [ 0.449006] fb0: VESA VGA frame buffer device [ 0.541671] ata1: SATA ........... <cut> I ran fbdump and it failed with segfault, tried again and it works.... [79549.693603] fbdump[1010]: segfault at ffffffffea208140 ip 00000000004018b4 sp 00007fffea2080b0 error 6 in fbdump[400000+3000]
I've here also a report from our build service ... due to the start of udev service something happens with /dev/console which cause blogs to loose its master tty side on the current system console (aka /dev/console) ... it seems that /dev/console is remapped by the kernel from /dev/pts/0 back to /dev/tty1. This could be caused by loading e.g. KMS modules due the rules of the udev service. This seems to be very racy as older (and much slower) blogd does run not into this problem or if RUN_PARALLEL if set to "no" or if an older kernel is booted. Beside this due bug #677435 the new blogd close the initial file descriptor 0 on /dev/console after the ioctl TIOCCONS to use explicit the open master side of the /dev/pts/0 as file descriptor 0. This was necessary with kernel 2.3.37. Strange
The main difference between 2.6.37 and e.g. 2.6.34 in drivers/char/tty_io.c seems to be the locking change from (un)lock_kernel forward to (un)lock_tty. This may have open a race ... if e.g. some other operations like loading modules for the console output/input are not covedrd by (un)lock_tty. Greg? Do you know what this change may cause on side of the user space?
*** Bug 680332 has been marked as a duplicate of this bug. ***
(In reply to comment #14) > The main difference between 2.6.37 and e.g. 2.6.34 in drivers/char/tty_io.c > seems to be the locking change from (un)lock_kernel forward to (un)lock_tty. > This may have open a race ... if e.g. some other operations like loading > modules for the console output/input are not covedrd by (un)lock_tty. > > Greg? Do you know what this change may cause on side of the user space? No I do not, sorry.
OK ... there must be a reason why the file descriptor is lost. For all reporters of this bug: which module for graphic output is used for the virtual consoles? lsmod | grep -E 'nv|nouv|intel|vesa|radeon|i[289]|r128|ati|drm'
A machine with exacly the same symptoms was upgraded from 11.3 to 11.4 and uses the nouveau module: lsmod | grep -E 'nv|nouv|intel|vesa|radeon|i[289]|r128|ati|drm' cpufreq_conservative 11828 0 snd_hda_intel 28391 0 snd_hda_codec 108050 2 snd_hda_codec_realtek,snd_hda_intel snd_pcm 104468 2 snd_hda_intel,snd_hda_codec snd 84374 6 snd_hda_codec_realtek,snd_hda_intel,snd_hda_codec,snd_hwdep,snd_pcm,snd_timer snd_page_alloc 9569 2 snd_hda_intel,snd_pcm i2c_nforce2 6851 0 nouveau 613485 1 ttm 74373 1 nouveau drm_kms_helper 36694 1 nouveau drm 232019 3 nouveau,ttm,drm_kms_helper i2c_algo_bit 6246 1 nouveau video 15929 1 nouveau button 6829 1 nouveau The system is configured to boot into run level 3.
Also an upgrade from 11.3 to 11.4 but with radeon: lsmod | grep -E 'nv|nouv|intel|vesa|radeon|i[289]|r128|ati|drm' -> radeon 912755 4 ttm 53631 1 radeon drm_kms_helper 26338 1 radeon drm 174214 6 radeon,ttm,drm_kms_helper i2c_algo_bit 4928 1 radeon
lsmod | grep -E 'nv|nouv|intel|vesa|radeon|i[289]|r128|ati|drm' snd_hda_intel 28391 4 snd_hda_codec 108050 2 snd_hda_codec_realtek,snd_hda_intel snd_pcm 104468 4 snd_pcm_oss,snd_hda_intel,snd_hda_codec snd 84374 16 snd_pcm_oss,snd_mixer_oss,snd_seq,snd_seq_device,snd_hda_codec_realtek,snd_hda_intel,snd_hda_codec,snd_hwdep,snd_pcm,snd_timer snd_page_alloc 9569 2 snd_hda_intel,snd_pcm cpufreq_conservative 11828 0 nvidia 10284430 52 kvm_intel 53490 0 kvm 374471 1 kvm_intel lirc_i2c 7921 0 lirc_dev 16018 1 lirc_i2c i2c_i801 10920 0 i2c_algo_bit 6246 1 ivtv
> lsmod | grep -E 'nv|nouv|intel|vesa|radeon|i[289]|r128|ati|drm' cpufreq_conservative 10080 0 snd_hda_intel 25072 2 snd_hda_codec 95606 2 snd_hda_codec_analog,snd_hda_intel i2c_i801 9448 0 snd_pcm 88249 3 snd_pcm_oss,snd_hda_intel,snd_hda_codec snd 65966 18 snd_pcm_oss,snd_mixer_oss,snd_seq,snd_hda_codec_analog,snd_hda_intel,thinkpad_acpi,snd_hda_codec,snd_hwdep,hdj_mod,snd_pcm,snd_timer,snd_rawmidi,snd_seq_device snd_page_alloc 8025 2 snd_hda_intel,snd_pcm radeon 938783 0 ttm 63435 1 radeon drm_kms_helper 32655 1 radeon drm 189801 3 radeon,ttm,drm_kms_helper i2c_algo_bit 5338 1 radeon
# grep fb boot.msg <4>[ 0.197390] ACPI: SSDT bfef1fb8 00085 (v01 PmRef Cpu1Cst 00000100 INTL 20050513) <6>[ 0.605295] vesafb: framebuffer at 0xd8000000, mapped to 0xf7a80000, using 5775k, total 16384k <6>[ 0.605303] vesafb: mode is 1400x1050x16, linelength=2816, pages=4 <6>[ 0.605307] vesafb: protected mode interface info at c000:b18a <6>[ 0.605311] vesafb: pmi: set display start = c00cb212, set palette = c00cb2ce <6>[ 0.605316] vesafb: scrolling: redraw <6>[ 0.605320] vesafb: Truecolor: size=0:5:6:5, shift=0:11:5:0 <6>[ 0.834259] fb0: VESA VGA frame buffer device service fbset start <notice -- Mar 16 20:30:49.257198000> service acpid done<notice -- Mar 16 20:30:49.258405000> service fbset done<notice -- Mar 16 20:30:49.258508000> service kbd start
(In reply to comment #13) > I've here also a report from our build service ... due to the > start of udev service something happens with /dev/console which > cause blogs to loose its master tty side on the current system > console (aka /dev/console) ... it seems that /dev/console is > remapped by the kernel from /dev/pts/0 back to /dev/tty1. > > This could be caused by loading e.g. KMS modules due the rules > of the udev service. This seems to be very racy as older (and > much slower) blogd does run not into this problem or if > RUN_PARALLEL if set to "no" or if an older kernel is booted. > > Beside this due bug #677435 the new blogd close the initial > file descriptor 0 on /dev/console after the ioctl TIOCCONS > to use explicit the open master side of the /dev/pts/0 > as file descriptor 0. This was necessary with kernel 2.3.37. > > Strange I have the issue even booting with the nomodeset in the kernel options every time.
(In reply to comment #23) This does not change the fact that something is destroying the pty master slave pair used by the blogd as system console. Facts are: The kernel is 2.6.37 and it does ont happen with 2.6.34. It happens during start of udev, that is that udev prepossess its rules found at /etc/udev/rules.d and /lib/udev/rules.d The problem is that due to the error exit blogd does not show the last few messages.
Created attachment 420653 [details] /sbin/blogd for 11.4-x86_64 Please on all with this problem: install the binary as /sbin/blogd do not forget to run chmod 755 /sbin/blogd after installation. Then please boot and see what happens. This version of blogd now uses some exit handlers to throw out the loggin ring buffer and tries to restore the old system console. A screen shot and/or the last few lines of /var/log/boot.msg may help to identify the cause aka the expanded udev rule,
Are there any results available?
Just installed sysvinit 2.88-37.43.1 and sysvinit-tools 2.88-37.43.1 from main update repo on my x86_64 system - but this problem still exists.
(In reply to comment #27) A screen shot please and the last few line of /var/log/boot.msg ... otherwise I've to guess whats happen ;)
Created attachment 420760 [details] Bootup after update to sysvinit(-tools) v2.88-37.43.1 from main update repo
Created attachment 420761 [details] Bootup after update to sysvinit(-tools) v2.88-37.43.1 from main update repo, with AppArmor disabled Why chmod 755? After a zypper in -f sysvinit-tools the file /sbin/blogd had the permissions 700 ! Nevertheless, I tried it with both permissions modes and had the same results. I also tried it a second time with AppArmor disabled (just had that idea - it's that boot-screen you can see on the second picture). But it didn't worked either. Just this time it initialized my FritzCard ISDN-Controller and the "random number generator" after the blogd error. But I remember that it initialized some services of runlevel 5 some on some startup tests I did before a few days. So it doesn't seem to be a certain point on startup, when blogd kills the runlevel initialization! My boot.msg: ... <6>[ 9.517406] em28xx driver loaded <6>[ 9.552242] em28xx-audio.c: probing for em28x1 non standard usbaudio <6>[ 9.552312] em28xx-audio.c: Copyright (C) 2006 Markus Rechberger <6>[ 9.552468] Em28xx: Initialized (Em28xx Audio Extension) extension <6>[ 9.666632] usbcore: registered new interface driver hdj_mod <4>[ 9.680674] zl10353_read_register: readreg error (reg=127, ret==-19) <3>[ 9.680790] em28xx #0: /2: dvb frontend not attached. Can't attach xc3028 <6>[ 9.680864] Em28xx: Initialized (Em28xx dvb Extension) extension <6>[ 9.759913] tvp5150 2-005c: tvp5150am1 detected. <6>[ 9.945596] Adding 4192928k swap on /dev/sda5. Priority:-1 extents:1 across:4192928k <4>[ 10.541152] device-mapper: ioctl: device doesn't appear to be in the dev hash table. <6>[ 10.805352] kjournald starting. Commit interval 15 seconds <6>[ 10.805595] EXT3-fs (dm-2): using internal journal <6>[ 10.805672] EXT3-fs (dm-2): mounted filesystem with ordered data mode <6>[ 10.852203] fuse init (API version 7.15) Kernel logging (ksyslog) stopped. Kernel log daemon terminating. Boot logging started on /dev/tty1(/dev/console) at Tue Mar 22 20:02:40 2011 Trying manual resume from /dev/disk/by-uuid/7f41bc7c-0132-4a1a-b3de-45d5b3a2fd58 Invoking userspace resume from /dev/disk/by-uuid/7f41bc7c-0132-4a1a-b3de-45d5b3a2fd58 resume: libgcrypt version: 1.4.6 Trying manual resume from /dev/disk/by-uuid/7f41bc7c-0132-4a1a-b3de-45d5b3a2fd58 Invoking in-kernel resume from /dev/disk/by-uuid/7f41bc7c-0132-4a1a-b3de-45d5b3a2fd58 Waiting for device /dev/disk/by-uuid/ea23481c-7061-4b71-9fc0-d18af2829156 to appear: ok fsck from util-linux 2.19 [/sbin/fsck.ext3 (1) -- /] fsck.ext3 -a -C0 /dev/sda1 root: clean, 410618/1640160 files, 3286852/6554512 blocks fsck succeeded. Mounting root device read-write. Mounting root /dev/disk/by-uuid/ea23481c-7061-4b71-9fc0-d18af2829156 mount -o rw,acl,user_xattr -t ext3 /dev/disk/by-uuid/ea23481c-7061-4b71-9fc0-d18af2829156 /root
Werner, please could you provide also 32bit version of that binary, co i can post my results? Thanks Gfs
Created attachment 420809 [details] boot.msg with blodg from comment 25
(In reply to comment #29) Moritz? You may try out the /bin/blogd binary of attachment #420653 [details] and for this you may consider to check out if the installed binary is executable.
Created attachment 420851 [details] /sbin/blogd for 11.4-i586
Created attachment 420880 [details] The last ca. 200 lines of my /var/log/boot.msg
I am the original reporter of this problem. With sysvinit-tools 2.88-40.1 (February 2011) I can not reproduce the problem, my system starts (apparently) correctly both with sysconfig RUN_PARALLEL=yes and RUN_PARALLEL=no. I get the message "blogd:Can not read from fd 0: Input/output error" which is mentioned in report #680332 on startup however, at least with RUN_PARALLEL=no. KMS is disabled on my system via the sysconfig variable NO_KMS_IN_INITRD. I have attached a snippet (last lines) of my /var/log/boot.msg anyway (RUN_PARALLEL=yes).
Are there any users of a serial console? If yes make sure that in /boot/grub/menu.lst the order of the consoles is: console=ttyS0,38400n8 console=tty0 compare with https://bugzilla.novell.com/show_bug.cgi?id=672450#c42
I don't have a serial console in use and no such line in menu.lst. So that's probably not the problem.
No serial console in use here either. If you make a 32-bit blogd binary available, I will be happy to collect some data for you.
Created attachment 421023 [details] Bootup using blogd from comment 33, giving debug output Startet my system, using the blogd binary from comment 33. It told this at startup (before crashing): blogd: system console stolen at line 263! There are also some services of runlevel 5, reported to be started. But stuff like the keyboard layout and xdm are still not coming up. I think it's just like that time without AppArmor I mentioned in comment 30. This is the boot.msg of the startup. It differs a lot from the ones I posted before: <6>[ 9.699710] em28xx driver loaded <6>[ 9.728564] em28xx-audio.c: probing for em28x1 non standard usbaudio <6>[ 9.728634] em28xx-audio.c: Copyright (C) 2006 Markus Rechberger <6>[ 9.728791] Em28xx: Initialized (Em28xx Audio Extension) extension <6>[ 9.826299] usbcore: registered new interface driver hdj_mod <4>[ 9.857446] zl10353_read_register: readreg error (reg=127, ret==-19) <3>[ 9.857574] em28xx #0: /2: dvb frontend not attached. Can't attach xc3028 <6>[ 9.857649] Em28xx: Initialized (Em28xx dvb Extension) extension <6>[ 9.971919] Adding 4192928k swap on /dev/sda5. Priority:-1 extents:1 across:4192928k <4>[ 11.402149] device-mapper: ioctl: device doesn't appear to be in the dev hash table. <6>[ 11.653129] kjournald starting. Commit interval 15 seconds <6>[ 11.653379] EXT3-fs (dm-2): using internal journal <6>[ 11.653458] EXT3-fs (dm-2): mounted filesystem with ordered data mode <6>[ 11.695266] fuse init (API version 7.15) Kernel logging (ksyslog) stopped. Kernel log daemon terminating. Boot logging started on /dev/tty1(/dev/console) at Wed Mar 23 23:41:13 2011 Trying manual resume from /dev/disk/by-uuid/7f41bc7c-0132-4a1a-b3de-45d5b3a2fd58 Invoking userspace resume from /dev/disk/by-uuid/7f41bc7c-0132-4a1a-b3de-45d5b3a2fd58 resume: libgcrypt version: 1.4.6 Trying manual resume from /dev/disk/by-uuid/7f41bc7c-0132-4a1a-b3de-45d5b3a2fd58 Invoking in-kernel resume from /dev/disk/by-uuid/7f41bc7c-0132-4a1a-b3de-45d5b3a2fd58 Waiting for device /dev/disk/by-uuid/ea23481c-7061-4b71-9fc0-d18af2829156 to appear: ok fsck from util-linux 2.19 [/sbin/fsck.ext3 (1) -- /] fsck.ext3 -a -C0 /dev/sda1 root: clean, 410807/1640160 files, 3292590/6554512 blocks fsck succeeded. Mounting root device read-write. Mounting root /dev/disk/by-uuid/ea23481c-7061-4b71-9fc0-d18af2829156 mount -o rw,acl,user_xattr -t ext3 /dev/disk/by-uuid/ea23481c-7061-4b71-9fc0-d18af2829156 /root Boot logging started on /dev/tty1(/dev/console) at Wed Mar 23 23:41:14 2011 <notice -- Mar 23 23:41:14.801570000> service boot.open-iscsi start <notice -- Mar 23 23:41:14.802051000> service boot.startpreload start <notice -- Mar 23 23:41:14.923896000> service boot.open-iscsi done <notice -- Mar 23 23:41:15.305419000> service boot.startpreload done done <notice -- Mar 23 23:41:15.305563000> service boot.udev start Starting udevd: done Loading drivers, configuring devices: ok done <notice -- Mar 23 23:41:20.993209000> service boot.udev done<notice -- Mar 23 23:41:20.993262000> service boot.loadmodules start <notice -- Mar 23 23:41:20.993451000> service boot.rootfsck start Loading required kernel modules <notice -- Mar 23 23:41:21.33630000> service boot.loadmodules done doneActivating swap-devices in /etc/fstab... done <notice -- Mar 23 23:41:21.75595000> service boot.rootfsck done <notice -- Mar 23 23:41:21.75649000> service boot.clock start Set System Time to the current Hardware ClockWed Mar 23 23:41:22 2011 -0.937639 seconds done <notice -- Mar 23 23:41:22.101446000> service boot.clock done <notice -- Mar 23 23:41:22.101649000> service boot.device-mapper start <notice -- Mar 23 23:41:22.101856000> service boot.localnet start <notice -- Mar 23 23:41:22.102081000> service boot.md start <notice -- Mar 23 23:41:22.122073000> service boot.device-mapper done Activating device mapper...done <notice -- Mar 23 23:41:22.200253000> service boot.md done Starting MD Raid unused <notice -- Mar 23 23:41:22.200322000> service boot.dmraid start Setting up hostname 'Trada'done Setting up loopback interface done <notice -- Mar 23 23:41:22.292413000> service boot.localnet done Waiting for udev to settle... Activating dmraid... RAID set "pdc_bebjhhcbgg" was activated RAID set "pdc_bebjhhcbgg" was not activated done <notice -- Mar 23 23:41:22.506217000> service boot.dmraid done<notice -- Mar 23 23:41:22.506270000> service boot.localfs start Waiting for /dev/mapper/pdc_bebjhhcbgg_part2 ok Checking file systems... fsck from util-linux 2.19 donedone Mounting local file systems... /dev/mapper/pdc_bebjhhcbgg_part2 on /home type ext3 (rw,acl,user_xattr) done <notice -- Mar 23 23:41:23.830644000> service boot.localfs done <notice -- Mar 23 23:41:23.830844000> service boot.cleanup start <notice -- Mar 23 23:41:23.831069000> service boot.cycle start <notice -- Mar 23 23:41:23.831300000> service boot.klog start<notice -- Mar 23 23:41:23.831543000> service boot.proc start <notice -- Mar 23 23:41:23.831747000> service boot.swap start <notice -- Mar 23 23:41:23.832009000> service boot.udev_retry start <notice -- Mar 23 23:41:23.832219000> service setserial start Finish udev device configuration: done <notice -- Mar 23 23:41:23.845530000> service boot.udev_retry done <notice -- Mar 23 23:41:23.856193000> service boot.proc done <notice -- Mar 23 23:41:23.935903000> service boot.cycle done <notice -- Mar 23 23:41:24.24902000> service setserial done Configuring serial ports... /dev/ttyS0 at 0x03f8 (irq = 4) is a 16550A Configured serial ports done <notice -- Mar 23 23:41:24.25250000> service boot.sysctl start Activating remaining swap-devices in /etc/fstab... done <notice -- Mar 23 23:41:24.47184000> service boot.swap done<notice -- Mar 23 23:41:24.47231000> service boot.ldconfig start <notice -- Mar 23 23:41:24.52415000> service boot.ldconfig done <notice -- Mar 23 23:41:24.88527000> service boot.sysctl done Setting current sysctl status from /etc/sysctl.confdone <notice -- Mar 23 23:41:24.88614000> service boot.ipconfig start <notice -- Mar 23 23:41:24.146122000> service boot.ipconfig done Enabling syn flood protectiondone Disabling IP forwardingdone Disabling IPv6 forwardingdone Disabling IPv6 privacydone <notice -- Mar 23 23:41:24.396063000> killproc: kill(422,29) Creating /var/log/boot.msg done <notice -- Mar 23 23:41:24.396691000> service boot.klog done ATTENTION: You have modified /etc/resolv.conf. Leaving it untouched... You can find my version in /etc/resolv.conf.netconfig ... <notice -- Mar 23 23:41:24.686620000> service boot.cleanup done System Boot Control: The system has been set up Skipped features: boot.open-iscsi boot.md boot.cycle System Boot Control: Running /etc/init.d/boot.local done <notice -- Mar 23 23:41:24.808635000> killproc: kill(422,3) Boot logging started on /dev/tty1(/dev/console) at Wed Mar 23 23:41:24 2011 Master Resource Control: previous runlevel: N, switching to runlevel: 5 Master Resource Control: Running /etc/init.d/before.local done <notice -- Mar 23 23:41:24.909509000> service earlysyslog start <notice -- Mar 23 23:41:24.910129000> service dbus start <notice -- Mar 23 23:41:24.910346000> service SuSEfirewall2_init start <notice -- Mar 23 23:41:24.910545000> service isdn start <notice -- Mar 23 23:41:24.910764000> service acpid start <notice -- Mar 23 23:41:24.910986000> service fbset start <notice -- Mar 23 23:41:24.911196000> service cpufreq start <notice -- Mar 23 23:41:24.911404000> service microcode.ctl start <notice -- Mar 23 23:41:24.911613000> service random start
Created attachment 421076 [details] /var/log/boot.msg The same result here: "blogd: system console stolen at line 263!" on console Please see attached /var/log/boot.msg. There's no serial console attached.
It is getting more weird: When shutting my system down and rebooting, it (again) does not boot correctly. So I have to correct my comment from yesterday: I still have the problem. I tested the blogd binary from comment 25 too, and also I updated my system to current opensuse-factory (sysvinit-tools 2.88+41.1). I do not think I have any serial console users. KMS is still disabled. With the blogd from comment 25 I see the message "blogd: system console stolen at line 263!" on boot. What follows are the last lines of /var/log/boot.msg with blogd from comment 25, RUN_PARALLEL=yes: Kernel logging (ksyslog) stopped. Kernel log daemon terminating. Boot logging started on /dev/tty1(/dev/console) at Thu Mar 24 10:52:14 2011 Waiting for device /dev/disk/by-id/ata-SAMSUNG_HD501LJ_S0MUJ13P103327-part1 to appear: ok fsck from util-linux 2.19 [/sbin/fsck.reiserfs (1) -- /] fsck.reiserfs -a /dev/sda1 Reiserfs super block in block 16 on 0x801 of format 3.6 with standard journal Blocks (total/free): 104858240/8110037 by 4096 bytes Filesystem is clean File system has been mounted 31 times without being checked. Checking now. Replaying journal: Replaying journal: Done. Reiserfs journal '/dev/sda1' in blocks [18..8211]: 0 transactions replayed File system has been mounted 31 times without being checked. Checking now. Checking internal tree.. ^H\^H|^H/^H-^H\^H|^H/^H-^H\^H|^H/^H-^H\^H|^H/^H-^H\^H|^H/^H-^H\^H|^H/^H-^H\^H|^H/^H-^H\^H|^H/^H-^H\^H|^H/^H-^H\^H|^H/^H-^H\^H|^H/^H-^H\^Hfinished fsck succeeded. Mounting root device read-write. Mounting root /dev/disk/by-id/ata-SAMSUNG_HD501LJ_S0MUJ13P103327-part1 mount -o rw,acl,user_xattr -t reiserfs /dev/disk/by-id/ata-SAMSUNG_HD501LJ_S0MUJ13P103327-part1 /root Boot logging started on /dev/tty1(/dev/console) at Thu Mar 24 10:52:41 2011 <notice -- Mar 24 10:52:41.333876000> service boot.udev start Starting udevd: udevd[362]: can not read '/etc/udev/rules.d/79-yast2-drivers.rules' done Loading drivers, configuring devices: <notice -- Mar 24 10:52:44.852742000> service boot.udev done done <notice -- Mar 24 10:52:44.866958000> service boot.loadmodules start Loading required kernel modules <notice -- Mar 24 10:52:44.894864000> service boot.loadmodules done done <notice -- Mar 24 10:52:44.894936000> service boot.rootfsck start Activating swap-devices in /etc/fstab... done <notice -- Mar 24 10:52:44.956223000> service boot.rootfsck done <notice -- Mar 24 10:52:44.971682000> service boot.device-mapper start <notice -- Mar 24 10:52:44.980917000> service boot.localnet start <notice -- Mar 24 10:52:45.91494000> service boot.md start <notice -- Mar 24 10:52:45.91839000> service boot.device-mapper done Activating device mapper...done <notice -- Mar 24 10:52:45.236903000> service boot.md done Starting MD Raid unused <notice -- Mar 24 10:52:45.245470000> service boot.lvm start <notice -- Mar 24 10:52:45.288346000> service boot.localnet done Setting up hostname 'linux-5rjk'done Setting up loopback interface done Waiting for udev to settle... Scanning for LVM volume groups... Reading all physical volumes. This may take a while... No volume groups found No volume groups found Activating LVM volume groups... No volume groups found <notice -- Mar 24 10:52:45.676538000> service boot.lvm done done <notice -- Mar 24 10:52:45.676611000> service boot.localfs start Checking file systems... fsck from util-linux 2.19 donedone Mounting local file systems... /dev/sda3 on /srv type reiserfs (rw) done <notice -- Mar 24 10:52:52.240523000> service boot.localfs done <notice -- Mar 24 10:52:52.257308000> service boot.crypto start Activating crypto devices using /etc/crypttab ... unused ...unused <notice -- Mar 24 10:52:52.385400000> service boot.crypto done <notice -- Mar 24 10:52:52.411705000> service boot.cleanup start <notice -- Mar 24 10:52:52.430253000> service boot.cycle start <notice -- Mar 24 10:52:52.430573000> service boot.klog start <notice -- Mar 24 10:52:52.450722000> service boot.proc start <notice -- Mar 24 10:52:52.451167000> service boot.swap start <notice -- Mar 24 10:52:52.464257000> service boot.udev_retry start <notice -- Mar 24 10:52:52.464595000> service boot.proc done <notice -- Mar 24 10:52:52.464662000> service setserial start <notice -- Mar 24 10:52:52.472333000> service boot.udev_retry done Finish udev device configuration: done <notice -- Mar 24 10:52:52.592416000> service boot.cycle done <notice -- Mar 24 10:52:52.720509000> service boot.swap done Activating remaining swap-devices in /etc/fstab... done <notice -- Mar 24 10:52:52.785748000> service setserial done Configuring serial ports... /dev/ttyS0 at 0x03f8 (irq = 4) is a 16550A Configured serial ports done <notice -- Mar 24 10:52:53.142588000> service boot.ldconfig start <notice -- Mar 24 10:52:53.232578000> service boot.sysctl start <notice -- Mar 24 10:52:53.356518000> service boot.ldconfig done <notice -- Mar 24 10:52:53.366287000> service boot.sysctl done Setting current sysctl status from /etc/sysctl.confdone <notice -- Mar 24 10:52:53.407063000> service boot.ipconfig start <notice -- Mar 24 10:52:53.415359000> service boot.ipconfig done Enabling syn flood protectiondone Disabling IP forwardingdone Disabling IPv6 forwardingdone Disabling IPv6 privacydone <notice -- Mar 24 10:52:54.21748000> killproc: kill(348,29) Creating /var/log/boot.msg done <notice -- Mar 24 10:52:54.22348000> service boot.klog done <notice -- Mar 24 10:52:55.468758000> service boot.cleanup done System Boot Control: The system has been set up Skipped features: boot.md boot.cycle System Boot Control: Running /etc/init.d/boot.local done <notice -- Mar 24 10:52:55.494498000> killproc: kill(348,3) Boot logging started on /dev/tty1(/dev/console) at Thu Mar 24 10:52:55 2011 Master Resource Control: previous runlevel: N, switching to runlevel: 5 Master Resource Control: Running /etc/init.d/before.local done <notice -- Mar 24 10:52:55.842585000> service earlysyslog start
Created attachment 421125 [details] /sbin/blogd for 11.4-i586
Created attachment 421127 [details] /sbin/blogd for 11.4-x86_64
Just solved two bugs: the usage of ttyname(3) in initrd which has changed in glibc of 11.4 and also a loop caused by writing data back to /dev/tty0 ... on systems with /proc/consoles this can not happen (kernel 2.6.38) but with 11.4 and kernel 2.6.37. Please test out the attachments: for 11.4-i586 use attachment #421125 [details] for 11.4-x86_64 use attachment #421127 [details] thanks.
When using attachment 421127 [details] I could restart my computer now three times successfully, with RUN_PARALLEL=yes. And I do not get the message "blogd: system console stolen at line 263!" anymore.
I tried it 2 times and no problems so far (x86_64). The bug occurred every time with the original blogd. Very good, thanks.
No problems so far for x86_64 after two reboot cycles. Appreciate your good work. Many thanks.
*** Bug 672450 has been marked as a duplicate of this bug. ***
IMHO a SWAMPID for an update is required.
Just want to confirm. I updated a i586 box from 11.2 to 11.4. Booting with keyboard works. Booting without keyboard does end up in "blogd:Can not read from fd 0: Input/output error". Using blogd attachment #421125 [details] solved the problem.
Yes, we should do an update (sysvinit-tools +1).
*** Bug 665720 has been marked as a duplicate of this bug. ***
Those attachments of the comments #421125 and #421127 work well on my x86_64 pc and my i586 notebook! No more problems! Thanks!
I'm bug 665720 reporter. In runlevel 3 boot I just replaced /sbin/blogd from 11.4 release with attachment 421125 [details], restored parallel boot to yes and rebooted to runlevel 3. Reported skipped services were cpufreq, nfs & rng-tools. Previous boot.msg is 76889 bytes. Current boot.msg is 80604 bytes. 4 straight boots without failures reported in bug 665720. :-)
Apparently 4 boots wasn't enough. On boots 6 and now 8 init simply stopped on "Starting syslog services", and CAD failed to finish rebooting after switch to runlevel 6 reported. :-(
Sometimes booting eventually completes after long delays after entering runlevel 3 at places like Starting acpid & Starting D-Bus daemon. Other times the delays are for practical purposes infinite, 3-5 minutes or more, and it's quicker to use the reset button to start over and wait on fscks than passively wait for a current delay to pass. None of these problems manifest on the same system booting 11.2.
(In reply to comment #57) This seems not to be related to the problem discussed within this bug. My guess is more dbus and ConsoleKit. For a try please disable dbus for next boot insserv -fr dbus ... as this is not a solution we have to investigate this precisely. Beside this with a working blogd you may have a look into /var/log/boot.msg you can compare the time stamps in the <notice> entries for the dbus service.
(In reply to comment #58) > This seems not to be related to the problem discussed within this bug. I filed bug 665720 duped to this. Maybe that should be reopened or a new bug filed instead of polluting this further with OT comments? Further response to comment 58 in that bug.
This bug (642289) was mentioned in https://build.opensuse.org/request/show/65683
(In reply to comment #60) I'm aware because I'm the author ;) and yes I'm awaiting that the request will now go into normal update.
The SWAMPID for this issue is 39939. This issue was rated as low. Please submit fixed packages until 2011-05-02. Also create a patchinfo file using this link: https://swamp.suse.de/webswamp/wf/39939
Accepted as update.
see https://rudin.suse.de:8894/request/show/65683 patchinfo done
needinfo done, also submitted...
*** Bug 681825 has been marked as a duplicate of this bug. ***
Update released for: sysvinit, sysvinit-debuginfo, sysvinit-debugsource, sysvinit-tools, sysvinit-tools-debuginfo Products: openSUSE 11.4 (debug, i586, x86_64)
*** Bug 688015 has been marked as a duplicate of this bug. ***
I just updated an other system from 11.2 to 11.4. After the update booting failed with error blogd: system console stolen at line 266! I have the latest sysvinit-tools-2.88-37.45.1.i586. I also tried attachment #421125 [details]. In this case booting went a little bit further but still hangs. As a workaround I just have disabled blogd with mv /sbin/blogd /sbin/blogd.disabled I think I'm not the only one seeing this problem: http://www.opensuse-forum.de/probleme-nach-aktualisierung-auf-11-4-dsdt-sysctl-conf-blogd-system-einrichten-verwalten/allgemeines-f17/t5282-f44/ Is this bug really resolved?
The problem should be fixed if the changelog of sysvinit-tools has following entry: rpm -q --changelog sysvinit-tools | head -n 4 * Fri Mar 25 2011 werner@suse.de - Fix triggered endless loop in blogd (bnc#642289) * Writing on tty0 caused blogd to re-read its own messages * The usage of ttyname(3) on /dev/console can fail I've run several test on several systems including with and without serial console. You may try out the version of sysvinit-tools for factory, see http://download.opensuse.org/repositories/Base:/System/openSUSE_11.4/ and report if this does work for you
Still not working for me. I have several other system updating from 11.3 or 11.2 to 11.4 were the problem was fixed with bnc#642289. But not on this system. I have latest sysvinit-tools from 11.4 rpm -q --changelog sysvinit-tools-2.88-37.45.1.i586 | head -n 4 * Fri Mar 25 2011 werner@suse.de - Fix triggered endless loop in blogd (bnc#642289) * Writing on tty0 caused blogd to re-read its own messages * The usage of ttyname(3) on /dev/console can fail and I tried http://download.opensuse.org/repositories/Base:/System/openSUSE_11.4/i586/sysvinit-tools-2.88+-162.1.i586.rpm BTW. In contrast to the other systems I have, this system was update from 10.3 to 11.2 and now to 11.4. Don't know, if this makes any difference.
Do have executed: mkinitrd after update? Guess: no ... but you should otherwise blogd in initrd remains the old one.
I did it no by intention. But for testing I installed kernel-pae in addition to kernel-default. But the problem still remained. Now I did mkinitrd for kernel-default and the problem is still there.
I'm not able to reproduce. Does RUN_PARALLEL="no" in /etc/sysconfig/boot help?
With RUN_PARALLEL="no" I still see blogd: system console stolen at line 266! but booting goes one. So this is "another workaround". Unfortunately, I can also not reproduce it on other systems...
Hmmm ... please show the type of /dev/console and /dev/tty0 by using ls -l /dev/console /dev/tty0 and an other question: do you use systemd or sysvinit? Also when does this message happen? Before enering an runlevel or before this step? Also what is that last message before the system console is stolen away?
This problem is still not solved for me. I did fresh installation using the 64 bit DVD => same problem I did a zypper dup using the repos installed using the automatic install => same problem I then re installed by doing a zypper in -f sysvinit and sysvinit-tools => same problem note : I checked the changelogs - see below - these rpms have the correct fixes! linux-xlle:/space1 # zypper in -f sysvinit Loading repository data... Reading installed packages... Forcing installation of 'sysvinit-2.88-37.45.1.x86_64' from repository 'Updates for openSUSE 11.4 11.4-0'. Resolving package dependencies... The following package is going to be reinstalled: sysvinit 1 package to reinstall. Overall download size: 67.0 KiB. No additional space will be used or freed after the operation. Continue? [y/n/?] (y): Retrieving package sysvinit-2.88-37.45.1.x86_64 (1/1), 67.0 KiB (108.0 KiB unpacked) Retrieving: sysvinit-2.88-37.45.1.x86_64.rpm [done (0 B/s)] Installing: sysvinit-2.88-37.45.1 [done] linux-xlle:/space1 # rpm -qa | grep sysv | rpm --changelog | head linux-xlle:/space1 # rpm -qa | grep sysv | rpm --changelog | head linux-xlle:/space1 # rpm -qa | grep sysv sysvinit-2.88-37.45.1.x86_64 sysvinit-tools-2.88-37.45.1.x86_64 linux-xlle:/space1 # rpm -q sysvinit-2.88-37.45.1.x86_64 --changelog | head * Fri Mar 25 2011 werner@suse.de - Fix triggered endless loop in blogd (bnc#642289) * Writing on tty0 caused blogd to re-read its own messages * The usage of ttyname(3) on /dev/console can fail * Thu Mar 10 2011 werner@suse.de - Make blogd work together with kernel from 11.4 even on a serial system console (bnc#672450) * Thu Mar 03 2011 werner@suse.de linux-xlle:/space1 # linux-xlle:/space1 # zypper in - f sysvinit-tools Loading repository data... Reading installed packages... 'sysvinit-tools' is already installed. No update candidate for 'sysvinit-tools-2.88-37.45.1.x86_64'. The highest available version is already installed. 'f' not found in package names. Trying capabilities. No provider of 'f' found. Resolving package dependencies... Problem: This request will break your system! glibc-2.11.3-12.15.1.x86_64 requires filesystem, but this requirement cannot be provided Solution 1: ignore the warning of a broken system (requires:glibc) Solution 2: keep filesystem-11.4-11.14.1.x86_64 Solution 3: break glibc by ignoring some of its dependencies Choose from above solutions by number or cancel [1/2/3/c] (c): linux-xlle:/space1 # zypper in -f sysvinit-tools Loading repository data... Reading installed packages... Forcing installation of 'sysvinit-tools-2.88-37.45.1.x86_64' from repository 'Updates for openSUSE 11.4 11.4-0'. Resolving package dependencies... The following package is going to be reinstalled: sysvinit-tools 1 package to reinstall. Overall download size: 169.0 KiB. No additional space will be used or freed after the operation. Continue? [y/n/?] (y): Retrieving package sysvinit-tools-2.88-37.45.1.x86_64 (1/1), 169.0 KiB (538.0 KiB unpacked) Retrieving: sysvinit-tools-2.88-37.45.1.x86_64.rpm [done (0 B/s)] Installing: sysvinit-tools-2.88-37.45.1 [done] Additional rpm output: Scanning scripts ... Resolve dependencies ... Install symlinks in /lib/mkinitrd/setup ... Install symlinks in /lib/mkinitrd/boot ... linux-xlle:/space1 # rpm -q sysvinit-tools-2.88-37.45.1.x86_64 --changelog | head * Fri Mar 25 2011 werner@suse.de - Fix triggered endless loop in blogd (bnc#642289) * Writing on tty0 caused blogd to re-read its own messages * The usage of ttyname(3) on /dev/console can fail * Thu Mar 10 2011 werner@suse.de - Make blogd work together with kernel from 11.4 even on a serial system console (bnc#672450) * Thu Mar 03 2011 werner@suse.de
I downloaded the version of sysvinit and sysvinit-tools version from factory as per Urs Beyerle suggestion using this url http://download.opensuse.org/repositories/Base:/System/openSUSE_11.4/ First remark - still no change - I have the same problem still Second remark - IMHO the fix has been applied to the factory rpms! Check for yourself below zypper in -f ./sysvinit-tools-2.88+-162.1.x86_64.rpm Loading repository data... Reading installed packages... Forcing installation of 'sysvinit-tools-2.88+-162.1.x86_64' from repository 'Plain RPM files cache'. Resolving package dependencies... The following package is going to be upgraded: sysvinit-tools The following package is going to change vendor: sysvinit-tools openSUSE -> obs://build.opensuse.org/Base 1 package to upgrade, 1 to change vendor. Overall download size: 174.0 KiB. After the operation, additional 8.0 KiB will be used. Continue? [y/n/?] (y): Retrieving package sysvinit-tools-2.88+-162.1.x86_64 (1/1), 174.0 KiB (546.0 KiB unpacked) Retrieving package sysvinit-tools-2.88+-162.1.x86_64 (1/1), 174.0 KiB (546.0 KiB unpacked) Installing: sysvinit-tools-2.88+-162.1 [done] Additional rpm output: warning: /var/cache/zypp/packages/_tmpRPMcache_/sysvinit-tools-2.88+-162.1.x86_64.rpm: Header V3 DSA/SHA1 Signature, key ID 5d053b4f: NOKEY Scanning scripts ... Resolve dependencies ... Install symlinks in /lib/mkinitrd/setup ... Install symlinks in /lib/mkinitrd/boot ... Scanning scripts ... Resolve dependencies ... Install symlinks in /lib/mkinitrd/setup ... Install symlinks in /lib/mkinitrd/boot ... There are some running programs that use files deleted by recent upgrade. You may wish to restart some of them. Run 'zypper ps' to list these programs. linux-xlle:/space1 # zypper in -f ./sysvinit-2.88+-162.1.x86_64.rpm Loading repository data... Reading installed packages... Forcing installation of 'sysvinit-2.88+-162.1.x86_64' from repository 'Plain RPM files cache'. Resolving package dependencies... The following package is going to be upgraded: sysvinit The following package is going to change vendor: sysvinit openSUSE -> obs://build.opensuse.org/Base 1 package to upgrade, 1 to change vendor. Overall download size: 69.0 KiB. No additional space will be used or freed after the operation. Continue? [y/n/?] (y): Retrieving package sysvinit-2.88+-162.1.x86_64 (1/1), 69.0 KiB (108.0 KiB unpacked) Retrieving package sysvinit-2.88+-162.1.x86_64 (1/1), 69.0 KiB (108.0 KiB unpacked) Installing: sysvinit-2.88+-162.1 [done] Additional rpm output: warning: /var/cache/zypp/packages/_tmpRPMcache_/sysvinit-2.88+-162.1.x86_64.rpm: Header V3 DSA/SHA1 Signature, key ID 5d053b4f: NOKEY There are some running programs that use files deleted by recent upgrade. You may wish to restart some of them. Run 'zypper ps' to list these programs. linux-xlle:/space1 # linux-xlle:/etc/init.d # rpm -qa | grep sysv sysvinit-tools-2.88+-162.1.x86_64 sysvinit-2.88+-162.1.x86_64 linux-xlle:/etc/init.d # rpm -q sysvinit-tools-2.88+-162.1.x86_64 --changelog | head * Fri Apr 15 2011 werner@suse.de - Fix exit code of checkproc in case of an existing pid file without running process (bnc#687547) * Wed Apr 06 2011 werner@suse.de - Fix bug in killproc that is do not stop searching for a match if a mountpoint does not match, reported by Friedrich Haubensak. * Fri Apr 01 2011 werner@suse.de - Let sulogin respect device on the comman line as well as the linux-xlle:/etc/init.d # linux-xlle:/etc/init.d # rpm -q sysvinit-2.88+-162.1.x86_64 --changelog | head * Fri Apr 15 2011 werner@suse.de - Fix exit code of checkproc in case of an existing pid file without running process (bnc#687547) * Wed Apr 06 2011 werner@suse.de - Fix bug in killproc that is do not stop searching for a match if a mountpoint does not match, reported by Friedrich Haubensak. * Fri Apr 01 2011 werner@suse.de - Let sulogin respect device on the comman line as well as the
Here is the additional info from my affected system: ls -l /dev/console /dev/tty0 crw------- 1 root root 5, 1 Apr 18 15:04 /dev/console crw--w---- 1 root tty 4, 0 Apr 18 15:04 /dev/tty0 I use sysvinit. My default runlevel is 3. It just happens when entering runlevel 3: INIT: Entering runlevel: 3 Boot logging started on /dev/tty1(/dev/console) at ... Master Resource Control: previous runlevel: N, switching to runlevel:3 Master Resource Control: Runnning /etc/init.d/before.local done blogd: system console stolen at line 266! blogd: Can not read from fd 0: Input/output error
# Concerning the type of /dev/console and /dev/tty0 by using ls -l /dev/console /dev/tty0 # My setup is as follows linux-xlle:/space1 # ls -l /dev/console /dev/tty0 crw------- 1 root root 5, 1 Apr 18 19:01 /dev/console crw--w---- 1 root root 4, 0 Apr 18 19:01 /dev/tty0 My default runlevel is 5 It is a default automatic install.
Why has this issue been marked as "resolved" ? With one particular system here running openSUSE 11.4 I hit this issue on every boot.
Please find a resume of my tests and findings to date below ######## Fresh install using the openSUSE-11.4-DVD-x86_64.iso - using the AUTOMATIC CONFIGURATION option when installing - network connection is up and running ######## Linux M3A78 2.6.37.1-1.2-desktop #1 SMP PREEMPT 2011-02-21 10:34:10 +0100 x86_64 x86_64 x86_64 GNU/Linux When I reboot after a FRESH INSTALL using the AUTOMATIC CONFIGURATION option I get the following error message blogd: system console stolen at line 266! ####################################################### Analysing what versions of sysvinit* are installed.... Found sysvinit-tools-2.88-37.40.1.x86_64 ! Reading the sysvinit-tools-2.88-37.40.1.x86_64 changelog... * Tue Feb 15 2011 werner@suse.de - New killproc version 2.19: bug fix update * Thu Feb 10 2011 werner@suse.de - Support the socket forwarding of systemd (bnc#656104) * Tue Feb 08 2011 werner@suse.de - Make real device comparision in killproc/checkproc to fix bnc#644171 - Also make ignore mode in checkproc work - Enhance mkill to work on root fs and ignore kernel threads ####################################################### Found sysvinit-2.88-37.40.1.x86_64 ! Reading the sysvinit-2.88-37.40.1.x86_64 changelog... * Tue Feb 15 2011 werner@suse.de - New killproc version 2.19: bug fix update * Thu Feb 10 2011 werner@suse.de - Support the socket forwarding of systemd (bnc#656104) * Tue Feb 08 2011 werner@suse.de - Make real device comparision in killproc/checkproc to fix bnc#644171 - Also make ignore mode in checkproc work - Enhance mkill to work on root fs and ignore kernel threads ####################################################### ####################################################### M3A78:/space1 # zypper lr # | Alias | Name | Enabled | Refresh --+----------------------------------+----------------------------------+---------+-------- 1 | Updates-for-openSUSE-11.4-11.4-0 | Updates for openSUSE 11.4 11.4-0 | Yes | Yes 2 | openSUSE-11.4-11.4-0 | openSUSE-11.4-11.4-0 | Yes | No 3 | repo-debug | openSUSE-11.4-Debug | No | Yes 4 | repo-debug-update | openSUSE-11.4-Update-Debug | No | Yes 5 | repo-non-oss | openSUSE-11.4-Non-Oss | Yes | Yes 6 | repo-oss | openSUSE-11.4-Oss | Yes | Yes 7 | repo-source | openSUSE-11.4-Source | No | Yes ######## Install the lastest versions of the sysvinit and sysvinit-tools - check the header to know what the latest changes are ######## M3A78:/space1 # zypper in -f sysvinit-tools Retrieving repository 'Updates for openSUSE 11.4 11.4-0' metadata [done] Building repository 'Updates for openSUSE 11.4 11.4-0' cache [done] Retrieving repository 'openSUSE-11.4-Non-Oss' metadata [done] Building repository 'openSUSE-11.4-Non-Oss' cache [done] Retrieving repository 'openSUSE-11.4-Oss' metadata [done] Building repository 'openSUSE-11.4-Oss' cache [done] Loading repository data... Reading installed packages... Forcing installation of 'sysvinit-tools-2.88-37.45.1.x86_64' from repository 'Updates for openSUSE 11.4 11.4-0'. Resolving package dependencies... The following package is going to be upgraded: sysvinit-tools 1 package to upgrade. Overall download size: 169.0 KiB. No additional space will be used or freed after the operation. Continue? [y/n/?] (y): Retrieving package sysvinit-tools-2.88-37.45.1.x86_64 (1/1), 169.0 KiB (538.0 KiB unpacked) Retrieving delta: ./rpm/x86_64/sysvinit-tools-2.88-37.40.1_37.45.1.x86_64.delta.rpm, 57.0 KiB Retrieving: sysvinit-tools-2.88-37.40.1_37.45.1.x86_64.delta.rpm [done] Applying delta: ./sysvinit-tools-2.88-37.40.1_37.45.1.x86_64.delta.rpm [done] Installing: sysvinit-tools-2.88-37.45.1 [done] Additional rpm output: Scanning scripts ... Resolve dependencies ... Install symlinks in /lib/mkinitrd/setup ... Install symlinks in /lib/mkinitrd/boot ... Scanning scripts ... Resolve dependencies ... Install symlinks in /lib/mkinitrd/setup ... Install symlinks in /lib/mkinitrd/boot ... M3A78:/space1 # zypper in -f sysvinit Loading repository data... Reading installed packages... Forcing installation of 'sysvinit-2.88-37.45.1.x86_64' from repository 'Updates for openSUSE 11.4 11.4-0'. Resolving package dependencies... The following package is going to be upgraded: sysvinit 1 package to upgrade. Overall download size: 67.0 KiB. No additional space will be used or freed after the operation. Continue? [y/n/?] (y): Retrieving package sysvinit-2.88-37.45.1.x86_64 (1/1), 67.0 KiB (108.0 KiB unpacked) Retrieving delta: ./rpm/x86_64/sysvinit-2.88-37.40.1_37.45.1.x86_64.delta.rpm, 24.0 KiB Retrieving: sysvinit-2.88-37.40.1_37.45.1.x86_64.delta.rpm [done (94.2 KiB/s)] Applying delta: ./sysvinit-2.88-37.40.1_37.45.1.x86_64.delta.rpm [done] Installing: sysvinit-2.88-37.45.1 [done] ####################################################### Linux M3A78 2.6.37.1-1.2-desktop #1 SMP PREEMPT 2011-02-21 10:34:10 +0100 x86_64 x86_64 x86_64 GNU/Linux Analysing what versions of sysvinit* are installed.... Found sysvinit-2.88-37.45.1.x86_64 ! Reading the sysvinit-2.88-37.45.1.x86_64 changelog... * Fri Mar 25 2011 werner@suse.de - Fix triggered endless loop in blogd (bnc#642289) * Writing on tty0 caused blogd to re-read its own messages * The usage of ttyname(3) on /dev/console can fail * Thu Mar 10 2011 werner@suse.de - Make blogd work together with kernel from 11.4 even on a serial system console (bnc#672450) * Thu Mar 03 2011 werner@suse.de ####################################################### Found sysvinit-tools-2.88-37.45.1.x86_64 ! Reading the sysvinit-tools-2.88-37.45.1.x86_64 changelog... * Fri Mar 25 2011 werner@suse.de - Fix triggered endless loop in blogd (bnc#642289) * Writing on tty0 caused blogd to re-read its own messages * The usage of ttyname(3) on /dev/console can fail * Thu Mar 10 2011 werner@suse.de - Make blogd work together with kernel from 11.4 even on a serial system console (bnc#672450) * Thu Mar 03 2011 werner@suse.de ####################################################### ####################################################### When I reboot after a FRESH INSTALL using the AUTOMATIC CONFIGURATION option and the lastest versions of the sysvinit and sysvinit-tools I ***still get the following error message*** blogd: system console stolen at line 266!
(In reply to comment #81) Because I'm not able to reproduce. The question is: What is the cause of having a blogd with a stolen system console?
(In reply to comment #83) > Because I'm not able to reproduce. The question is: What is the > cause of having a blogd with a stolen system console? There must still be some bug present in either the sysvinit process or blogd, maybe a race condition. I administer two openSUSE 11.4 systems, and on both systems blogd fails frequently causing the startup process to abort - at least if RUN_PARALLEL is set to "yes". I have not yet seen this happening with RUN_PARALLEL set to "no". If blogd fails it either complains about a stolen console or that it cannot read from fd 0. Both systems are updated daily.
Created attachment 425592 [details] rc.diff ro be applied to /etc/init.d/rc This small patch may help to detect what happens on the affected systems.
I've been putting all my boot problem comments in bug 665720 lately, but I just booted in order to apply the comment 85 patch, and encountered both "system console stolen at line 266!" and "Can not read from fd 0: Input/output error". Boot seemed to complete in otherwise apparently normal time and manner, but NUM was not on, and eth0 had no IP. What else failed to start I don't know how to tell, except that between RAID startup messages and runlevel 3 has been reached were only 17 line: /dev/md6 on /home... Finish udev device con... Setting current sysctlc st... Creating /var/log/boot.msg Activating remaining swap... System Boot Control: Running /etc/init.d/boot.local Boot logging started on /dev/tty1... Master Resource Control: previous runlevel: N... Master Resource Control: Running /etc/init.d/before.local blogd: system console stolen at line 266! blogd: Can not read from fd 0: Input/output error acpid: starting up with proc fs Starting acpid acpid: 2 rules loaded acpid: waitig for events: event logging is off Master Resource Control: Running /etc/init.d/after.local Master Resource Control: runlevel 3... Welcome to ... I applied the patch, but got a strace usage message after "...Running /etc/init.d/boot.local" instead of /tmp/blogd.strace.
In other words blogd ould already be died before strace
Just tried to remove any possible race from blogd, see latest sysvinit-tools from factory (sysvinit-tools-2.88+-164.1 and above). Should be visible at http://download.opensuse.org/repositories/Base:/System/openSUSE_11.4/ in a few minutes or an hour or so.
I downloaded and installed the 2 rpm which Werner Fink mentionned - see "Comment 88" above at 2011-04-19 14:39:28 CEST When I rebooted - I had the same problem Note - the error occurred at line 288 this time - before it stopped at line 266 Error message follows... blog: System console stolen at line 288 blog: Can not read from fd 0: Input/output error Starting syslog service If you look below you can see that the 2 new versions of sysvinit-tools-2.88+-165.1.x86_64 and sysvinit-2.88+-165.1.x86_64 are installed. ####################################################### Linux M3A78 2.6.37.1-1.2-desktop #1 SMP PREEMPT 2011-02-21 10:34:10 +0100 x86_64 x86_64 x86_64 GNU/Linux Analysing what versions of sysvinit* are installed.... Found sysvinit-tools-2.88+-165.1.x86_64 ! Reading the sysvinit-tools-2.88+-165.1.x86_64 changelog... * Tue Apr 19 2011 werner@suse.de - Correct shutdown messages of startpar send via blogd * Tue Apr 19 2011 werner@suse.de - Avoid possible races which can be happen if blogd sees a signal and will exit then (related to bnc#642289) * Fri Apr 15 2011 werner@suse.de - Fix exit code of checkproc in case of an existing pid file without running process (bnc#687547) ####################################################### Found sysvinit-2.88+-165.1.x86_64 ! Reading the sysvinit-2.88+-165.1.x86_64 changelog... * Tue Apr 19 2011 werner@suse.de - Correct shutdown messages of startpar send via blogd * Tue Apr 19 2011 werner@suse.de - Avoid possible races which can be happen if blogd sees a signal and will exit then (related to bnc#642289) * Fri Apr 15 2011 werner@suse.de - Fix exit code of checkproc in case of an existing pid file without running process (bnc#687547) ####################################################### #######################################################
Created attachment 425766 [details] boot.msg & boot.omsg from boots 9 & 10 I didn't notice "blog: System console stolen at line 288 & blog: Can not read from fd 0: Input/output error" until the 9th boot after installing 164.1 of i586 sysvinit & sysvinit-tools. So far, every boot since has proceeded to available login prompts. 56k boot.msg from boot 9 with those messages is 20k smaller than 76k boot.omsg. Boot #10 also produced those messages, much closer to the login prompt, left keyboard settings undone, network down, and me confused to find a 63k boot.omsg with a 58k boot.msg. Are boot.msg and/or boot.omsg getting written on shutdown? Boot #11 displayed no apparent errors, boot.omsg 62543, boot.msg 70393. Boot #12 no apparent errors, 75985 boot.omsg, 70530 boot.msg. Boot #13 no apparent errors, 76047 boot.omsg, 70366 boot.msg, stalled noticeably executing reboot, leaving at least one partition uncleanly umounted. Boot #14 75953 boot.omsg, 71169 boot.msg. Boot #15 76705 boot.omsg, 70354 boot.msg. Boot #16 75867 boot.omsg, 70449 boot.msg. Boot #18 was next occurrence of "blog: System console stolen at line 288 & blog: Can not read from fd 0: Input/output error", boot.omsg 76068, boot.msg 57861. Boot #19 63457 boot.omsg, 70410 boot.msg. Boot #20 stalled noticeably at shutdown, but 20 boots without failing to reach a login prompt seems to mean improvement has been made. To compare, same machine to 11.2 on /dev/md3, boot.omsg 67689, boot.msg 61726. Boot #2 boot.omsg 67130, boot.msg 61797. Boot #3 67201 boot.omsg, 61514 boot.msg. Boot #4 66913 boot.omsg, 61488 boot.msg. Boot #5 66893 boot.omsg, 61486 boot.msg. Smaller variation in sizes for 11.2.
Hmmm ... guess: One of the major changes in blogd and startpar was the usage of a raw pts/ptmx terminal line (no post processing no echo no canonical mode) to speed up the writing onto slave side and reading from the masters side. Maybe this could be the reason to see an EIO due a read on fast systems. Now let us test this with sysvinit-tools-2.88+-166.1 at http://download.opensuse.org/repositories/Base:/System/openSUSE_11.4/ in a few minutes or so
I downloaded and installed the the new versions of sysvinit-tools-2.88+-166.1.x86_64.rpm and sysvinit-2.88+-166.1.x86_64.rpm (see Comment 91) When I rebooted - I had the same problem Note - the error occurred at line 293 this time - last time it occurred at line 288 :) Error message follows... blog: System console stolen at line 288 blog: Can not read from fd 0: Input/output error Starting syslog service Below you can see the changelog of the 2 new versions of sysvinit-tools-2.88+-166.1.x86_64.rpm and sysvinit-2.88+-166.1.x86_64.rpm I installed ####################################################### Linux M3A78 2.6.37.1-1.2-desktop #1 SMP PREEMPT 2011-02-21 10:34:10 +0100 x86_64 x86_64 x86_64 GNU/Linux Analysing what versions of sysvinit* are installed.... Found sysvinit-tools-2.88+-166.1.x86_64 ! Reading the sysvinit-tools-2.88+-166.1.x86_64 changelog... * Wed Apr 20 2011 werner@suse.de - Aoid possible trouble due raw pts/ptmx terminal line in both blogd and startpar - Block SIGTTOU during tcsetattr(3) library call in both blogd and startpar - Replace select(2) with pselect(2) in blogd and ensure that the timeout structure will be reseted after a timeout * Tue Apr 19 2011 werner@suse.de - Correct shutdown messages of startpar send via blogd ####################################################### Found sysvinit-2.88+-166.1.x86_64 ! Reading the sysvinit-2.88+-166.1.x86_64 changelog... * Wed Apr 20 2011 werner@suse.de - Aoid possible trouble due raw pts/ptmx terminal line in both blogd and startpar - Block SIGTTOU during tcsetattr(3) library call in both blogd and startpar - Replace select(2) with pselect(2) in blogd and ensure that the timeout structure will be reseted after a timeout * Tue Apr 19 2011 werner@suse.de - Correct shutdown messages of startpar send via blogd ####################################################### Hope this helps
First boot with 2.88+-166.1 installed for me produced: blog: System console stolen at line 293! blog: Can not read from fd 0: Input/output error acpid: starting up with proc fs Starting acpid ... Starting kernel based NFS server: idmapd mountdrpc.mountd: svc_tli_create: could not open connection for udp6 rpc.mountd: svc_tli_create: could not open connection for tcp6 rpc.mountd: svc_tli_create: could not open connection for udp6 rpc.mountd: svc_tli_create: could not open connection for tcp6 rpc.mountd: svc_tli_create: could not open connection for udp6 rpc.mountd: svc_tli_create: could not open connection for tcp6 statd nfsdrpc.nfsd: unable to create inet6 TCP socket: errno 97 (Address family not supported by protocol) sm-notify ... Skipped services in runlevel 3: cpufreq rng-tools nfs Welcome to... boot.omsg 77607, boot.msg 59420 Boot #2 acpid service failed, boot.omsg 64592, boot.msg 72013. Boot #3 boot.omsg 77160, boot.msg 59423. Boot #4 blog: System console stolen at line 293! blog: Can not read from fd 0: Input/output error, 64594 boot.omsg, 59516 boot.msg. Boot #5 boot.omsg 64654, boot.msg 59469. Boot #6 boot.omsg 64676, boot.msg 72072. Boot #7 blog: System console stolen at line 293! blog: Can not read from fd 0: Input/output error, boot.omsg 77280, boot.msg 59420.
Please add the line killproc -KILL /sbin/blogd *before* /sbin/blogd $REDIRECT in /etc/init.d/rc ... maybe we see on heavily loaded systems a race that is that the former blogd is not stopped at all.
done Following: boot #1 65381 boot.omsg, 73637 boot.msg. Boot #2 79392 boot.omsg, 72570 boot.msg. Boot #3 78172 boot.omsg, 72687 boot.msg. Boot #4 78432 boot.omsg, 72672 boot.msg. Boot #5 78217 boot.omsg, 72558 boot.msg. Boot #6 78286 boot.omsg, 72742 boot.msg. Boot #7 78520 boot.omsg, 72811 boot.msg. Boot #8 78568 boot.omsg, 72804 boot.msg. Boot #9 78577 boot.omsg, 72838 boot.msg. Boot #10 78437 boot.omsg, 72831 boot.msg. Boot #11 78442 boot.omsg, 72755 boot.msg. Boot #12 78531 boot.omsg, 72886 boot.msg. No failure messages noticed on tty1 during inits. Each time when login prompt appeared on tty1, "Starting acpid" was either on first or second non-blank line on top of screen. Before its location was much less consistent.
Wow! In other words due to high terminal I/O the first blogd started in /etc/init.d/boot had not finished yet and the second stareted in /etc/init.d/rc gets its system console stolen from the first one.
Just submitted a fixed aaa_base for 11.4 (see obs request #68050) and also for factory. With this only one blogd is started in user space even on cold boot. This may require a further SWAMP ID to submit the fixed aaa_base but before set needinfo to maintenance@opensuse.org I'd like to see some testers, see https://build.opensuse.org/package/binaries?arch=x86_64&package=aaa_base&project=home%3AWernerFink%3Abranches%3AopenSUSE%3A11.4%3AUpdate%3ATest&repository=standard
Installed 2 comment 97 packages, then: boot# boot.msg boot.omsg comment 1 72671 78614 2 72670 78494 3 72712 78494 4 72724 78529 5 72733 78490 6 72672 78383 7 72675 78489 8 72671 78475 stalled shutdown, unclean umount 9 72536 78490 auto fsck on /dev/sda1 (on /disks/hda/boot) 10 72684 79305 11 72686 78330 12 72664 78344
(In reply to comment #98) You may have a look into the boot.msg respectively boot.omsg to see what has happend. Also the question is: does this also happen without blogd?
Beside this an auto fsck may happen due reached mount/umount count the question is what had caused the unclean umount (it is very unlikely that blogd had caused this).
(In reply to comment #99) > Also the question is: does this also happen without blogd? How do I test? rm sbin/blogd? insserv -d earlysyslog? Remove some package? Other? The shutdown pauses causing unclean umount on sda1 are random but common, once in somewhere around 7-16 boots. All my partitions have had 'tune2fs -c0 -i0 <device>' applied.
I did a fresh install this morning using the openSUSE-11.4-DVD-x86_64.iso - using the AUTOMATIC CONFIGURATION option when I installed - the network connection was up and running When the desktop / machine came up on the first soft boot - I did a zypper up - I installed the sysvinit-tools-2.88+-166.1.x86_64.rpm and sysvinit-2.88+-166.1.x86_64.rpm - see the changelogs below Reading the sysvinit-2.88-37.45.1.x86_64 changelog... * Fri Mar 25 2011 werner@suse.de - Fix triggered endless loop in blogd (bnc#642289) * Writing on tty0 caused blogd to re-read its own messages * The usage of ttyname(3) on /dev/console can fail * Thu Mar 10 2011 werner@suse.de - Make blogd work together with kernel from 11.4 even on a serial system console (bnc#672450) * Thu Mar 03 2011 werner@suse.de Found sysvinit-tools-2.88-37.45.1.x86_64 ! Reading the sysvinit-tools-2.88-37.45.1.x86_64 changelog... * Fri Mar 25 2011 werner@suse.de - Fix triggered endless loop in blogd (bnc#642289) * Writing on tty0 caused blogd to re-read its own messages * The usage of ttyname(3) on /dev/console can fail * Thu Mar 10 2011 werner@suse.de - Make blogd work together with kernel from 11.4 even on a serial system console (bnc#672450) * Thu Mar 03 2011 werner@suse.de When I rebooted - I had the blog erorr type problems Ex: System console stolen at line 293! blog: Can not read from fd 0: Input/output error The changelogs of the aaa_base and aaa_base-extras I was using when I was experiencing the problems are below. ####################################################### Reading the aaa_base-11.4-54.62.1.x86_64 changelog... * Fri Mar 11 2011 werner@suse.de - Avoid to trap into execute escapes (bnc#678827) * Tue Feb 22 2011 werner@suse.de - Work around colon as breaking character in tab completion - Allow arguments of command done by sudo to complete (bnc#673663) * Fri Feb 18 2011 werner@suse.de - Fix minimal support for the old fashion rc service links ####################################################### Reading the aaa_base-extras-11.4-54.62.1.x86_64 changelog... * Fri Mar 11 2011 werner@suse.de - Avoid to trap into execute escapes (bnc#678827) * Tue Feb 22 2011 werner@suse.de - Work around colon as breaking character in tab completion - Allow arguments of command done by sudo to complete (bnc#673663) * Fri Feb 18 2011 werner@suse.de - Fix minimal support for the old fashion rc service links ####################################################### ####################################################### Tonight I downloaded and installed the two new versions of aaa_base-11.4-54.66.1.x86_64.rpm and aaa_base-extras-11.4-54.66.1.x86_64.rpm I have sucessfully rebooted 3 times with no problems. This is good news! I will try to test this on another machine over the weekend. The changelogs of the aaa_base and aaa_base-extras when I was running when I successfully booted are below. ####################################################### Linux M3A78 2.6.37.1-1.2-desktop #1 SMP PREEMPT 2011-02-21 10:34:10 +0100 x86_64 x86_64 x86_64 GNU/Linux Analysing what versions of sysvinit* are installed.... Found aaa_base-11.4-54.66.1.x86_64 ! Reading the aaa_base-11.4-54.66.1.x86_64 changelog... * Thu Apr 21 2011 werner@suse.de - Start blogd only once at boot (related to bnc#642289) * Mon Mar 21 2011 werner@suse.de - Avoid waiting on bind mounts in boot.localfs - Be aware in refresh_initrd that modules used in initrd may use options in the /etc/modprobe.d/ files * Fri Mar 11 2011 werner@suse.de - Avoid to trap into execute escapes (bnc#678827) ####################################################### Found aaa_base-extras-11.4-54.66.1.x86_64 ! Reading the aaa_base-extras-11.4-54.66.1.x86_64 changelog... * Thu Apr 21 2011 werner@suse.de - Start blogd only once at boot (related to bnc#642289) * Mon Mar 21 2011 werner@suse.de - Avoid waiting on bind mounts in boot.localfs - Be aware in refresh_initrd that modules used in initrd may use options in the /etc/modprobe.d/ files * Fri Mar 11 2011 werner@suse.de - Avoid to trap into execute escapes (bnc#678827) ####################################################### ####################################################### Hope this helps.
I just ran another test as per "comment 102" - the only difference this time was that I installed openSUSE-11.4-DVD-i586.iso. The test machine was a Thinkpad T60 The test procedure followed was as comment 102" The problems encountered using the openSUSE-11.4-DVD-i586.iso were also the same (as per comment 102). By that I mean that when I rebooted - I had the blog error type problems Ex: System console stolen at line 293! blog: Can not read from fd 0: Input/output error After confirming this behavior I then installed sysvinit-tools-2.88+-166.1.i586.rpm sysvinit-2.88+-166.1.i586.rpm Result I still had a blog error I reboot again and I installed the new aaa_base-11.4-54.66.1.i586.rpm aaa_base-extras-11.4-54.66.1.i586.rpm Result No more blog errors Conclusion All 4 rpm must be installed to solve blog type errors My comments follows... The command "hdparm -t /dev/sda" on my T60 returns 101.99 MB/sec My other test machine had a raptor (10000 rpm) hard drive. Both hard drives are fast. My test observations (as well as all my v11.4 experiences to date) lead me to believe that anyone with a fast hard drive (where fast is ~100 MB/sec or more) will encounter blog type errors when installing v114 using either of the default (32 and 64 bit) 11.4 DVD regardless of whether he runs the "update" during the installation process or not. If my observations are true, then the 4 rpms above should be added to the v11.4 update queue ASP. Regards James
Created attachment 426318 [details] zypper log attached zypper log showing: - zypper up post fresh install plus - the standard packman packages which I always install Note : I always use the same script when I install the packman rpms
I tested my hypothese that the blogd issue only affects fast hard drives today by only pulling out an old HP nc6120 laptop. # machine specs HP nc6120 laptop vendor_id : GenuineIntel cpu family : 6 model : 13 model name : Intel(R) Pentium(R) M processor 1.73GHz Hard Drive Model is: SAMSUNG HM160HC /dev/sda: SAMSUNG HM160HC: 41°C Timing buffered disk reads: 194 MB in 3.03 seconds = 64.11 MB/sec ############# I did a fresh install this morning using the openSUSE-11.4-DVD-i586.iso - and the AUTOMATIC CONFIGURATION option when I installed - the network connection was up and running When the desktop / machine came up on the first soft boot - I did a zypper The pc booted OK the first time without the blogd issue. I added the Packman repo and added a few packages as per usual The blogd issue appeared next boot Once I installed the 4 rpms below the blogd issue disappeared. ####################################################### zypper -f ./sysvinit-2.88+-166.1.i586.rpm zypper in -f ./sysvinit-2.88+-166.1.i586.rpm zypper in -f ./aaa_base-11.4-54.66.1.i586.rpm zypper in -f ./sysvinit-tools-2.88+-166.1.i586.rpm ####################################################### Analysing what versions of sysvinit* are installed.... ####################################################### Found sysvinit-tools-2.88+-166.1.i586 ! Reading the sysvinit-tools-2.88+-166.1.i586 changelog... * Wed Apr 20 2011 werner@suse.de - Aoid possible trouble due raw pts/ptmx terminal line in both blogd and startpar - Block SIGTTOU during tcsetattr(3) library call in both blogd and startpar - Replace select(2) with pselect(2) in blogd and ensure that the timeout structure will be reseted after a timeout * Tue Apr 19 2011 werner@suse.de - Correct shutdown messages of startpar send via blogd Found sysvinit-2.88+-166.1.i586 ! Reading the sysvinit-2.88+-166.1.i586 changelog... * Wed Apr 20 2011 werner@suse.de - Aoid possible trouble due raw pts/ptmx terminal line in both blogd and startpar - Block SIGTTOU during tcsetattr(3) library call in both blogd and startpar - Replace select(2) with pselect(2) in blogd and ensure that the timeout structure will be reseted after a timeout * Tue Apr 19 2011 werner@suse.de - Correct shutdown messages of startpar send via blogd ####################################################### Analysing what versions of aaa_base* are installed.... ####################################################### Found aaa_base-extras-11.4-54.62.1.i586 ! Reading the aaa_base-extras-11.4-54.62.1.i586 changelog... * Fri Mar 11 2011 werner@suse.de - Avoid to trap into execute escapes (bnc#678827) * Tue Feb 22 2011 werner@suse.de - Work around colon as breaking character in tab completion - Allow arguments of command done by sudo to complete (bnc#673663) * Fri Feb 18 2011 werner@suse.de - Fix minimal support for the old fashion rc service links Reading the aaa_base-11.4-54.66.1.i586 changelog... * Thu Apr 21 2011 werner@suse.de - Start blogd only once at boot (related to bnc#642289) * Mon Mar 21 2011 werner@suse.de - Avoid waiting on bind mounts in boot.localfs - Be aware in refresh_initrd that modules used in initrd may use options in the /etc/modprobe.d/ files * Fri Mar 11 2011 werner@suse.de - Avoid to trap into execute escapes (bnc#678827) ####################################################### ####################################################### my zypper.log showing the Packman rpm which I installed is under comment 104
(In reply to comment #101) Simply by moving blogd away mv /sbin/blogd /sbin/blogd.back
This is an autogenerated message for OBS integration: This bug (642289) was mentioned in https://build.opensuse.org/request/show/68050
A further SWAMP ID is required to put the fixed aaa_base out.
lets do +1
swamp opened: https://swamp.suse.de/webswamp/swamp/template/DisplayWorkflow.vm/workflowid/40537 I also submitted a patchinfo for this and the bind mount bug.
*** Bug 691049 has been marked as a duplicate of this bug. ***
(In reply to comment #110) Correct patchinfo (it is only a 11.4 issue)
(In reply to comment #69) > I just updated an other system from 11.2 to 11.4. After the update booting > failed with error > > blogd: system console stolen at line 266! > > I have the latest sysvinit-tools-2.88-37.45.1.i586. > I also tried attachment #421125 [details]. In this case booting went a little bit further > but still hangs. > > As a workaround I just have disabled blogd with > > mv /sbin/blogd /sbin/blogd.disabled > > I think I'm not the only one seeing this problem: > http://www.opensuse-forum.de/probleme-nach-aktualisierung-auf-11-4-dsdt-sysctl-conf-blogd-system-einrichten-verwalten/allgemeines-f17/t5282-f44/ > > Is this bug really resolved? I've this issue too (after 11.2 to 11.4 upgrading). If apparmor is enabled, i've this issue(pls see attachment), if apparmor is disabled - my system is booting. I've any fresh updates installed at this time.
Created attachment 427866 [details] System boot screenshot (when apparmor is enabled)
aaa_base for 11.4 release 54.64.1 is out with the changelog entry * Thu Apr 21 2011 werner@suse.de - Start blogd only once at boot (related to bnc#642289) does this mean that apparmor can destroy blogd's terminal aka the temprary system console?
(In reply to comment #115) > aaa_base for 11.4 release 54.64.1 is out with the changelog entry > > * Thu Apr 21 2011 werner@suse.de > - Start blogd only once at boot (related to bnc#642289) > > does this mean that apparmor can destroy blogd's terminal aka > the temprary system console? now, i've installed aaa_base-11.5-305.1.i586.rpm with aaa_base-extras-11.5-305.1.i586.rpm from http://download.opensuse.org/repositories/Base:/System/openSUSE_11.4, and now, my PC can to boot with apparmor enabled.
perhaps apparmor causes timing changes that triggers it?
(In reply to comment #117) Exactly, as the blogd of 11.4 is listen a few ms after it has been signaled to stop. Simply to not to loose the last few messages on the system console (bug #593957 and bug #623460). Now this changes leads to the fact the if the first blogd, started in /etc/init.d/boot, may stop at the same time as the next blogd will be started in /etc/init.d/rc, the former can steal the system console from the fresh started blogd. Now with the new aaa_base only one blogd will be started and the problem is gone.
Update released for: aaa_base, aaa_base-debuginfo, aaa_base-debugsource, aaa_base-extras Products: openSUSE 11.4 (debug, i586, x86_64)
Fixed and verfied
This is an autogenerated message for OBS integration: This bug (642289) was mentioned in https://build.opensuse.org/request/show/70685 Factory / sysvinit
*** Bug 723220 has been marked as a duplicate of this bug. ***