http://fm.no-ip.com/Tmp/Linux/boot.log-f15r-t2240.txt http://fm.no-ip.com/Tmp/Linux/dmesg-f15r-t2240.txt http://fm.no-ip.com/Tmp/Linux/messages-f15r-t2240.txt
Can anyone tell me from those logs why booting into any runlevel beyond 1 simply stops, with no keyboard response possible at all, or in some cases only CAD works?
I tried a dist upgrade from F14 yesterday, and it hasn't booted completely into other than runlevel 1 since. CPU is P4celeron@2.4G on i845G, and X doesn't start successfully if I attempt booting directly to runlevel 5.
On Sat, Jan 01, 2011 at 03:12:01PM -0500, Felix Miata wrote:
http://fm.no-ip.com/Tmp/Linux/boot.log-f15r-t2240.txt http://fm.no-ip.com/Tmp/Linux/dmesg-f15r-t2240.txt http://fm.no-ip.com/Tmp/Linux/messages-f15r-t2240.txt
The last one comes back with "Access forbidden!".
Can anyone tell me from those logs why booting into any runlevel beyond 1 simply stops,
Apparently, from your boot.log, something gets stuck trying to configure network interfaces. Why this is the case you are the one in a position to investigate.
Either boot to a level 1, turn off whatever tries to configure eth0 and boot to a level 3, or go into an "interactive control of services" mode and skip whatever gets stuck. Once at a command line try to bring up an offending service (or services) manually and watch what goes haywire.
I tried a dist upgrade from F14 yesterday,
It is possible then that you bumped into some problems with systemd. Quite likely it would be truly good to know what really went wrong.
and X doesn't start successfully if I attempt booting directly to runlevel 5.
X requires as a minimum at least properly configured loopback network interface and a basic sanity of network parameters (in particular if you require an outside network access to resolve a host name then you are in trouble with a networking down). In any case do not worry about X before you can boot to a login prompt.
Michal
On 2011/01/01 15:18 (GMT-0700) Michal Jaegermann composed:
On Sat, Jan 01, 2011 at 03:12:01PM -0500, Felix Miata wrote:
http://fm.no-ip.com/Tmp/Linux/boot.log-f15r-t2240.txt http://fm.no-ip.com/Tmp/Linux/dmesg-f15r-t2240.txt http://fm.no-ip.com/Tmp/Linux/messages-f15r-t2240.txt
The last one comes back with "Access forbidden!".
I had neglected to change the default permissions on the original when copying to the web server dir. That's fixed now.
Can anyone tell me from those logs why booting into any runlevel beyond 1 simply stops,
Apparently, from your boot.log, something gets stuck trying to configure network interfaces. Why this is the case you are the one in a position to investigate.
Ordinarily I would, but I've never found the manual page explaining how to do with upstart the things I knew how to do with sysvinit. Investigate where?
Either boot to a level 1, turn off whatever tries to configure eth0 and boot to a level 3, or go into an "interactive control of services" mode and skip whatever gets stuck. Once at a command line try to bring up an offending service (or services) manually and watch what goes haywire.
I've got Rawhide running and yum updated further via chroot from Factory boot, but don't know what to look for as indicated above.
I tried a dist upgrade from F14 yesterday,
It is possible then that you bumped into some problems with systemd. Quite likely it would be truly good to know what really went wrong.
On Sat, Jan 01, 2011 at 07:11:14PM -0500, Felix Miata wrote:
On 2011/01/01 15:18 (GMT-0700) Michal Jaegermann composed:
Apparently, from your boot.log, something gets stuck trying to configure network interfaces. Why this is the case you are the one in a position to investigate.
Ordinarily I would, but I've never found the manual page explaining how to do with upstart the things I knew how to do with sysvinit.
You have the same service scripts but if you moved up from F14 to rawhide then you will be using systemd. Your /etc/inittab is likely replaced. Check explanatory comments there.
Investigate where?
Where it gets stuck. 'sh -x /etc/init.d/network start' looks like a good beginning and you are following up from this.
Either boot to a level 1, turn off whatever tries to configure eth0 and boot to a level 3, or go into an "interactive control of services" mode and skip whatever gets stuck. Once at a command line try to bring up an offending service (or services) manually and watch what goes haywire.
I've got Rawhide running and yum updated further via chroot from Factory boot, but don't know what to look for as indicated above.
If you booted to level 1 then 'chkconfig network off' and/or similar. If you are just booting "normally" then you need to drop 'rhgb quiet' from a boot command and when a user space starts to show up you have a message on yor screen "For an interactive startup press I" or something pretty close. It is pretty loud and clear.
Michal
On 2011/01/01 20:57 (GMT-0700) Michal Jaegermann composed:
On Sat, Jan 01, 2011 at 07:11:14PM -0500, Felix Miata wrote:
I've never found the manual page explaining how to do with upstart the things I knew how to do with sysvinit.
You have the same service scripts but if you moved up from F14 to rawhide then you will be using systemd. Your /etc/inittab is likely replaced. Check explanatory comments there.
They seem minimally useful. This is the entirety of inittab: # inittab is no longer used when using systemd. # # ADDING CONFIGURATION HERE WILL HAVE NO EFFECT ON YOUR SYSTEM. # # Non-SysV tasks for individual runlevels live in /lib/systemd/system/runlevelX.target.wants # # Ctrl-Alt-Delete is handled by /etc/systemd/system/ctrl-alt-del.target # # To set a default runlevel <X>, run: # # ln -s /lib/systemd/system/runlevel<X>.target /etc/systemd/system/default.target
Investigate where?
Where it gets stuck. 'sh -x /etc/init.d/network start' looks like a good beginning and you are following up from this.
If I boot to runlevel 1 and do that, network is up, Google is pingable.
On a normal boot to runlevel 3 it appears to proceed well beyond that. Maybe those logs are misleading? first line on screen after attempting runlevel 3 boot is: Starting Avahi mDNS/DNS-SD Stack... following is a bunch of LSB: lines, last of which is: Mount and umnount network filesystems.... Starting /etc/rc.local Compatibility... Starting Terminate Plymouth Boot Screen... Retrigger failed udev events Starting cups: Starting sshd: Mounting other filesystems: Starting NFS statd: Starting RPC idmapd: [42.171289] RPC: Registered udp transport module. ... starting rpcsvcgsse (via systemctl... Initializing OpenCT smart card... (some USB lines) Starting LSB: Daemon to access a smart c... Starting PC/SC smart c... Starting HAL Hardware Manager... [43.414506] Installing knfsd... Starting NFS services: Starting NFS daemon: [44.0996085] NFSD: Using /var/lib/nfs/v4recovery as the NFSv4 state recovery directory [44.311521] NFSD: starting 90-second grace period Starting NFS mountd: [OK] [44.708309] uhci_hcd 0000:00:1d.0: PCI INT A disabled [44.724072] ehci_hcd 0000:00:1d.7: PCI INT B disabled [OK] [44.712166] uhci_hcd 0000:00:1d.1: PCI INT C disabled [44.719190] uhci_hcd 0000:00:1d.2: PCI INT D disabled [OK]
From here, nothing more happens, no keyboard response, except to shift-PgUp or CAD.
From the former, I see following "eth0: link becomes ready" four lines of: grep: ifcfg-ifcfg-: No such file or directory, then: Bringing up interface ifcfg-: Starting LSB:...
Maybe that's about an unconfigured/unwanted/unneeded (onboard) eth1? eth0 is a configured PCI card.
The only instances of "failed" I see doing Shift-PgUps are: Starting Recreate Volatile Files and Directories failed. [33.849718] systemd[1]: Unit systemd-tempfiles-setup.service entered failed state. ...last write time is in the future...
until I get above mounting messages and see at least 9 lines of of invalid EDID....
then further up, 3 lines of failed to write '0' to /proc/sys/.../tables: no such...
I tried removing /etc/exports, /etc/nfs.conf and all lines containing NFS from /etc/fstab, but that didn't help either.
I've got Rawhide running and yum updated further via chroot from Factory boot, but don't know what to look for as indicated above.
If you booted to level 1 then 'chkconfig network off' and/or similar. If you are just booting "normally" then you need to drop 'rhgb quiet' from a boot command and when a user space starts to
I've been stripping all instances of rhgb quiet from grub.conf for as long as they've been appearing there.
show up you have a message on yor screen "For an interactive startup press I" or something pretty close. It is pretty loud and clear.
I've been seeing that for a lot of years, but I don't see it now. Maybe it's getting lost in mode switches/switching, or maybe it's that little bit of illegible navy text on black background most of the way through init? How far along should it proceed before I try?
On Sun, Jan 02, 2011 at 12:37:43PM -0500, Felix Miata wrote:
On 2011/01/01 20:57 (GMT-0700) Michal Jaegermann composed:
Investigate where?
Where it gets stuck. 'sh -x /etc/init.d/network start' looks like a good beginning and you are following up from this.
If I boot to runlevel 1 and do that, network is up, Google is pingable.
On a normal boot to runlevel 3 it appears to proceed well beyond that. Maybe those logs are misleading?
That is why it really up to you to find out what is wrong. Walk through you boot sequence step by step and find out where there are problems and why.
Starting NFS mountd: [OK] [44.708309] uhci_hcd 0000:00:1d.0: PCI INT A disabled [44.724072] ehci_hcd 0000:00:1d.7: PCI INT B disabled [OK] [44.712166] uhci_hcd 0000:00:1d.1: PCI INT C disabled [44.719190] uhci_hcd 0000:00:1d.2: PCI INT D disabled [OK]
From here, nothing more happens,
So what should be next? Check a listing of /etc/rc.d/rc3.d/S* to find what you configured. Are you sure that the previous step really finished? Answering such questions is much easier if you are running that explicitely instead of making it happen "by itself".
Michal
On 2011/01/02 11:37 (GMT-0700) Michal Jaegermann composed:
it really up to you to find out what is wrong. Walk through you boot sequence step by step and find out where there are problems and why.
So what should be next? Check a listing of /etc/rc.d/rc3.d/S* to find what you configured. Are you sure that the previous step really finished? Answering such questions is much easier if you are running that explicitely instead of making it happen "by itself".
Finally I figured out (I think) what you meant I should do: boot runlevel 1 cd /etc/rc.d/rc3.d/ (each) `./S* start` in turn
First error: bash: ./S50bluetooth: No such file or directory
There should be no bluetooth configured, as there is no such hardware on this 7 year old system.
Hmmm, no more errors through 99rc-local, and all exited.
Hmmm again. After some time: [900.089897] systemd[1]: systemd-tmpfiles-clean.service: main process exited, code=exited, status=1 [900.094000] systemd[1]: Unit systemd-tmpfiles-clean.service entered failed state.
After running all that, runlevel still shows S. What did I miss?
What if any significance is there to systemd-tmpfiles-clean.service going into failed state?
How & when do I get into interactive startup, since I don't any more see any message suggesting that option?
On Sun, Jan 02, 2011 at 04:53:49PM -0500, Felix Miata wrote:
First error: bash: ./S50bluetooth: No such file or directory
symlink -d /etc/rc.d/rc?.d/
There should be no bluetooth configured, as there is no such hardware on this 7 year old system.
You should have cleaned that up (as above).
You should also possibly do:
cd /etc/rc.d/init.d && \ for f in *; do /sbin/chkconfig $f resetpriorities; done
Hmmm, no more errors through 99rc-local, and all exited.
So something else is blocking you although this looks somewhat puzzling. Maybe systemd is indeed a culprit? Maybe something else? Missing getty processes? You can try to use SysRq key (check Google for docs if you do not know what I am talking about) but before attempting that you need to set kernel.sysrq = 1 in /etc/sysctl.conf before a boot attempt.
Hmmm again. After some time: [900.089897] systemd[1]: systemd-tmpfiles-clean.service: main process exited, code=exited, status=1 [900.094000] systemd[1]: Unit systemd-tmpfiles-clean.service entered failed state.
That is https://bugzilla.redhat.com/show_bug.cgi?id=662490 and it should not really affect what you doing beyond making a noise.
After running all that, runlevel still shows S. What did I miss?
That running various services does not change a boot level. Check 'man telinit' (and "SEE ALSO" section there).
How & when do I get into interactive startup,
See the previous replies but remember to put "3" into a boot line instead of "rhgb quiet".
Michal
On 2011/01/02 15:25 (GMT-0700) Michal Jaegermann composed:
On Sun, Jan 02, 2011 at 04:53:49PM -0500, Felix Miata wrote:
How & when do I get into interactive startup,
See the previous replies but remember to put "3" into a boot line instead of "rhgb quiet".
To be clear, as I alluded to in my previous replies, none of my (100+) OS installations has rhgb or quiet on any kernel line in any stanza I ever use, while nearly every kernel line among Grub stanzas contains a 3. I detest obfuscatory GUI startup. I read every word you wrote, and asked again only because I have no idea why it never shows up, other than maybe it's that illegible blue text I do notice, or it's getting lost among the screen clearing that occurs more than once during each boot. Should I just hold down the I key as soon as anything shows up on screen? If not, what landmarks among messages should I wait for? I tried random times, but I get either nothing at all, or just streams of iiiiiiii or IIIIIIII.
On Sun, Jan 02, 2011 at 06:37:34PM -0500, Felix Miata wrote:
On 2011/01/02 15:25 (GMT-0700) Michal Jaegermann composed:
On Sun, Jan 02, 2011 at 04:53:49PM -0500, Felix Miata wrote:
How & when do I get into interactive startup,
See the previous replies but remember to put "3" into a boot line instead of "rhgb quiet".
To be clear, as I alluded to in my previous replies, none of my (100+) OS installations has rhgb or quiet on any kernel line in any stanza I ever use,
Then I am at loss how to explain that you may miss "Press 'I' to enter interactive startup." message which shows up in a distro banner.
Michal
Michal Jaegermann (michal@harddata.com) said:
To be clear, as I alluded to in my previous replies, none of my (100+) OS installations has rhgb or quiet on any kernel line in any stanza I ever use,
Then I am at loss how to explain that you may miss "Press 'I' to enter interactive startup." message which shows up in a distro banner.
Because the message was removed in F-14/RHEL 6?
Bill
On 01/03/2011 11:55 AM, Bill Nottingham wrote:
Michal Jaegermann (michal@harddata.com) said:
To be clear, as I alluded to in my previous replies, none of my (100+) OS installations has rhgb or quiet on any kernel line in any stanza I ever use,
Then I am at loss how to explain that you may miss "Press 'I' to enter interactive startup." message which shows up in a distro banner.
Because the message was removed in F-14/RHEL 6?
Probably because that feature has been very unreliable, if it worked at all, for quite a few releases before that.
Adding "confirm" as a boot parameter will reliably get you into interactive startup. (Tried it just now in F-12 and F-14.)
On 2011/01/02 15:25 (GMT-0700) Michal Jaegermann composed:
On Sun, Jan 02, 2011 at 04:53:49PM -0500, Felix Miata wrote:
Hmmm, no more errors through 99rc-local, and all exited.
So something else is blocking you although this looks somewhat puzzling. Maybe systemd is indeed a culprit? Maybe something else? Missing getty processes? You can try to use SysRq key (check Google for docs if you do not know what I am talking about) but before attempting that you need to set kernel.sysrq = 1 in /etc/sysctl.conf before a boot attempt.
...
That running various services does not change a boot level. Check 'man telinit' (and "SEE ALSO" section there).
If after running all of S* start in rc3.d/ I do telinit 3, instead of getting a console with a login prompt (which announces Fedora 13 :-O ), KDM starts, and all other ttys are void of login prompts. I tried yum reinstall mingetty, but that didn't cause any apparent change.
This is S* in /etc/rc3.d/: lrwxrwxrwx 1 root root 16 Jan 1 15:27 S07iscsid -> ../init.d/iscsid lrwxrwxrwx 1 root root 17 Dec 31 22:20 S10network -> ../init.d/network lrwxrwxrwx 1 root root 16 Mar 17 2010 S11auditd -> ../init.d/auditd lrwxrwxrwx 1 root root 21 Dec 31 23:21 S11portreserve -> ../init.d/portreserve lrwxrwxrwx 1 root root 17 Dec 31 23:20 S12rsyslog -> ../init.d/rsyslog lrwxrwxrwx 1 root root 15 Jan 1 15:27 S13iscsi -> ../init.d/iscsi lrwxrwxrwx 1 root root 17 Mar 17 2010 S13rpcbind -> ../init.d/rpcbind lrwxrwxrwx 1 root root 19 Mar 17 2010 S15mdmonitor -> ../init.d/mdmonitor lrwxrwxrwx 1 root root 20 Dec 31 22:19 S22messagebus -> ../init.d/messagebus lrwxrwxrwx 1 root root 22 Dec 31 22:33 S24avahi-daemon -> ../init.d/avahi-daemon lrwxrwxrwx 1 root root 17 Dec 31 23:29 S24nfslock -> ../init.d/nfslock lrwxrwxrwx 1 root root 16 Jan 1 15:27 S24openct -> ../init.d/openct lrwxrwxrwx 1 root root 17 Dec 31 23:29 S24rpcgssd -> ../init.d/rpcgssd lrwxrwxrwx 1 root root 19 Dec 31 23:29 S24rpcidmapd -> ../init.d/rpcidmapd lrwxrwxrwx 1 root root 14 Jan 1 13:20 S25cups -> ../init.d/cups lrwxrwxrwx 1 root root 15 Dec 31 22:20 S25netfs -> ../init.d/netfs lrwxrwxrwx 1 root root 19 Mar 17 2010 S26haldaemon -> ../init.d/haldaemon lrwxrwxrwx 1 root root 19 Dec 31 22:20 S26udev-post -> ../init.d/udev-post lrwxrwxrwx 1 root root 15 Dec 31 23:08 S27pcscd -> ../init.d/pcscd lrwxrwxrwx 1 root root 13 Dec 31 23:09 S30nfs -> ../init.d/nfs lrwxrwxrwx 1 root root 14 Dec 31 23:19 S55sshd -> ../init.d/sshd lrwxrwxrwx 1 root root 14 Dec 31 23:08 S58ntpd -> ../init.d/ntpd lrwxrwxrwx 1 root root 15 Dec 31 22:33 S90crond -> ../init.d/crond lrwxrwxrwx 1 root root 11 Dec 31 22:20 S99rc-local -> ../rc.local
Shouldn't there be something in there starting gettys, like a fbset? RPM query found no installed fbset package, but yum install fbset (2.1.29.fc12 ???) didn't add anything to /etc/rc3.d/, or help anything I could tell.
On 2011/01/02 19:43 (GMT-0500) Felix Miata composed:
Shouldn't there be something in there starting gettys, like a fbset? RPM query found no installed fbset package, but yum install fbset (2.1.29.fc12 ???) didn't add anything to /etc/rc3.d/, or help anything I could tell.
Yum install fbterm directfb not only didn't help, kernel segfaults shortly after boot starts.
I'd now try a fresh install, but there are no Rawhide installation files on the mirrors that I can find except in development/14/ dated 29 October.
Michal Jaegermann michal@harddata.com wrote:
On Sun, Jan 02, 2011 at 04:53:49PM -0500, Felix Miata wrote:
First error: bash: ./S50bluetooth: No such file or directory
symlink -d /etc/rc.d/rc?.d/
There should be no bluetooth configured, as there is no such hardware on this 7 year old system.
You should have cleaned that up (as above).
You should also possibly do:
cd /etc/rc.d/init.d && \ for f in *; do /sbin/chkconfig $f resetpriorities; done
Hmmm, no more errors through 99rc-local, and all exited.
So something else is blocking you although this looks somewhat puzzling. Maybe systemd is indeed a culprit? Maybe something else? Missing getty processes?
Here (on an x86_64, Fedora rawhide) I have:
# systemctl status getty.service getty.service Loaded: error Active: inactive (dead)
Even so, a bunch of getty's show up here. On another (newer) machine there are no gettys at all, and fixing things by ctrl-alt-F2 doesn't lead anywhere.
On 2011/01/02 23:54 (GMT-0300) Horst H. von Brand composed:
Michal Jaegermannmichal@harddata.com wrote:
So something else is blocking you although this looks somewhat puzzling. Maybe systemd is indeed a culprit? Maybe something else? Missing getty processes?
Here (on an x86_64, Fedora rawhide) I have:
# systemctl status getty.service getty.service Loaded: error Active: inactive (dead)
Even so, a bunch of getty's show up here. On another (newer) machine there are no gettys at all, and fixing things by ctrl-alt-F2 doesn't lead anywhere.
https://bugzilla.redhat.com/show_bug.cgi?id=665890 :-(
On Sun, Jan 02, 2011 at 10:08:47PM -0500, Felix Miata wrote:
On 2011/01/02 23:54 (GMT-0300) Horst H. von Brand composed:
Michal Jaegermannmichal@harddata.com wrote:
So something else is blocking you although this looks somewhat puzzling. Maybe systemd is indeed a culprit? Maybe something else? Missing getty processes?
Here (on an x86_64, Fedora rawhide) I have:
# systemctl status getty.service getty.service Loaded: error Active: inactive (dead)
Even so, a bunch of getty's show up here. On another (newer) machine there are no gettys at all, and fixing things by ctrl-alt-F2 doesn't lead anywhere.
With a dead getty service you still should be able to reach that machine over a network as long as sshd is running. Then you may try to check what is going on.
Michal
On Sun, 2011-01-02 at 11:37 -0700, Michal Jaegermann wrote:
So what should be next? Check a listing of /etc/rc.d/rc3.d/S* to find what you configured. Are you sure that the previous step really finished? Answering such questions is much easier if you are running that explicitely instead of making it happen "by itself".
on rawhide, not everything is going to be there; systemd native scripts live elsewhere entirely. it may be best to use the systemd tools to investigate the boot process.