All reboot tests are made with console mode (CTRL + ALT + F1) and CTRL + ALT + SUPPR
Sometimes, reboot hungs on apache.
I had a problem with apache2, which doesn't want to pass because of there wasn't "ServerName localhost" options in /etc/apache2/apache2.conf.
Still not clean reboot :
Before, every computer, at every time hangs.
I added "reboot=pci" kernel option in /etc/default/grub, and now, if a computer was rebooted few minutes before a new reboot try, the problem comes 1 times on 10.
It hangs after all K** in rc6.d.
I noticed that, the difference between computers that reboot properly and those which do not is a log line :
"Rpcbind terminated on signal" : when this line apear, one second later, the computer reboots properly.
I googled it and saw that RPCbind has a report with NFS and that, that is "S20sendsigs" which send this signal.
I don't know where autofs is shutting down. So i thank that umountnfs.sh was executed before autofs closed.
So, I created a script sifrebootdefrozen which contains :
#!/bin/bash
service autofs stop
sleep 2
And place it in K02. When it is executed, i saw on screen "autofs already closed" or something like that. So that's not the problem.
If I wait for 30 minutes or more, a new problem appears, reboot hangs on black screen, only "acpid exiting" diplayed.
I tried every "acpi=" kernel options in /etc/default/grub. Still same problem, so i removed acpid (with its dependence acpi-support)
apt-get remove acpid (key words for googling)
I don't have acpid log line diplayed. Acpi is clearly not the problem too.
So i excluded every others parameters, i can now test kernels in a cleaner way.
Didn't try these kernels yet but, for posterity :
All reboot tests are made with console mode (CTRL + ALT + F1) and CTRL + ALT + SUPPR
Sometimes, reboot hungs on apache. apache2. conf.
I had a problem with apache2, which doesn't want to pass because of there wasn't "ServerName localhost" options in /etc/apache2/
Still not clean reboot :
Before, every computer, at every time hangs.
I added "reboot=pci" kernel option in /etc/default/grub, and now, if a computer was rebooted few minutes before a new reboot try, the problem comes 1 times on 10.
It hangs after all K** in rc6.d.
I noticed that, the difference between computers that reboot properly and those which do not is a log line :
"Rpcbind terminated on signal" : when this line apear, one second later, the computer reboots properly.
I googled it and saw that RPCbind has a report with NFS and that, that is "S20sendsigs" which send this signal.
Here my rc6.d dir :
lrwxrwxrwx 1 root root 21 avril 23 11:09 K01sifupdate64 -> ../init. d/sifupdate64 rozer -> ../init. d/sifrebootdefr ozer upgrades -> ../init. d/unattended- upgrades d/libnss- ldap dispatcher -> ../init. d/speech- dispatcher d/virtualbox d/x2goserver kernel- server -> ../init. d/nfs-kernel- server d/umountnfs. sh d/networking d/umountroot
lrwxrwxrwx 1 root root 27 avril 23 11:11 K02sifrebootdef
lrwxrwxrwx 1 root root 17 oct. 17 2013 K08tomcat7 -> ../init.d/tomcat7
lrwxrwxrwx 1 root root 17 oct. 17 2013 K09apache2 -> ../init.d/apache2
lrwxrwxrwx 1 root root 29 oct. 16 2013 K10unattended-
lrwxrwxrwx 1 root root 21 oct. 17 2013 K20libnss-ldap -> ../init.
lrwxrwxrwx 1 root root 14 oct. 17 2013 K20nscd -> ../init.d/nscd
lrwxrwxrwx 1 root root 15 oct. 17 2013 K20nslcd -> ../init.d/nslcd
lrwxrwxrwx 1 root root 27 oct. 16 2013 K20speech-
lrwxrwxrwx 1 root root 20 nov. 4 10:18 K20virtualbox -> ../init.
lrwxrwxrwx 1 root root 17 févr. 11 16:22 K20winbind -> ../init.d/winbind
lrwxrwxrwx 1 root root 20 janv. 20 14:27 K20x2goserver -> ../init.
lrwxrwxrwx 1 root root 18 oct. 17 2013 K79quotarpc -> ../init.d/quotarpc
lrwxrwxrwx 1 root root 27 oct. 17 2013 K80nfs-
lrwxrwxrwx 1 root root 15 oct. 17 2013 K85quota -> ../init.d/quota
-rw-r--r-- 1 root root 351 juil. 26 2012 README
lrwxrwxrwx 1 root root 18 oct. 16 2013 S20sendsigs -> ../init.d/sendsigs
lrwxrwxrwx 1 root root 17 oct. 16 2013 S30urandom -> ../init.d/urandom
lrwxrwxrwx 1 root root 22 oct. 16 2013 S31umountnfs.sh -> ../init.
lrwxrwxrwx 1 root root 20 oct. 16 2013 S35networking -> ../init.
lrwxrwxrwx 1 root root 18 oct. 16 2013 S40umountfs -> ../init.d/umountfs
lrwxrwxrwx 1 root root 20 oct. 16 2013 S60umountroot -> ../init.
lrwxrwxrwx 1 root root 16 oct. 16 2013 S90reboot -> ../init.d/reboot
I don't know where autofs is shutting down. So i thank that umountnfs.sh was executed before autofs closed.
So, I created a script sifrebootdefrozen which contains :
#!/bin/bash
service autofs stop
sleep 2
And place it in K02. When it is executed, i saw on screen "autofs already closed" or something like that. So that's not the problem.
If I wait for 30 minutes or more, a new problem appears, reboot hangs on black screen, only "acpid exiting" diplayed.
I tried every "acpi=" kernel options in /etc/default/grub. Still same problem, so i removed acpid (with its dependence acpi-support)
apt-get remove acpid (key words for googling)
I don't have acpid log line diplayed. Acpi is clearly not the problem too.
So i excluded every others parameters, i can now test kernels in a cleaner way.
Best regards.