the mentioned files are not being processed or setting are not applied. I don't know which package it belongs to Reproducible: Always Steps to Reproduce: 1. write net.ipv6.conf.all.use_tempaddr = 2 in /etc/sysctl.conf or /etc/sysctl.d/local.conf 2. reboot 3. cat /proc/sys/net/ipv6/conf/all/use_tempaddr 0 Expected Results: 2 kernel features are enabled
Please post your `emerge --info' output in a comment.
Created attachment 351574 [details] emerge --info
Created attachment 351578 [details] emerge --info 2nd pc
Could you investigate this further? Is sysctl simply not being run at startup? If so, does sysctl succeed when you run: sysctl net.ipv6.conf.all.use_tempaddr=2 in a shell?
(In reply to Jeroen Roovers from comment #4) > Could you investigate this further? Is sysctl simply not being run at > startup? If so, does sysctl succeed when you run: > sysctl net.ipv6.conf.all.use_tempaddr=2 > in a shell? Yes it succeeds and works next time I boot (on both PCs) but is reset to 0 on the boot that follows reading rc.log it seems that it isn't ran however sysctl | boot rc boot logging started at Sat Jun 22 01:11:38 2013 * Setting system clock using the hardware clock [UTC] ... hwclock: Cannot access the Hardware Clock via any known method. hwclock: Use the --debug option to see the details of our search for an access method. * Failed to set the system clock [ !! ] * Autoloaded 0 module(s) * Setting up the Logical Volume Manager ... No volume groups found No volume groups found No volume groups found [ ok ] * Checking local filesystems ... /dev/sdb1: clean, 495694/29483008 files, 5309136/117902848 blocks [ ok ] * Remounting root filesystem read/write ... [ ok ] * Remounting filesystems ... [ ok ] * Updating /etc/mtab ... [ ok ] * Activating swap devices ... [ ok ] * Mounting local filesystems ... [ ok ] * Configuring kernel parameters ... [ ok ] * Starting ZFS ... * Checking if zfs userspace tools present * Importing ZFS pools * Mounting ZFS filesystems * Exporting ZFS filesystems [ ok ] * Creating user login records ... [ ok ] * Cleaning /var/run ... [ ok ] * Wiping /tmp directory ... [ ok ] * Setting hostname to server ... [ ok ] * Setting terminal encoding [UTF-8] ... [ ok ] * Setting keyboard mode [UTF-8] ... [ ok ] * Loading key mappings [de] ... [ ok ] * Bringing up interface lo * 127.0.0.1/8 ... [ ok ] * Adding routes * 127.0.0.0/8 via 127.0.0.1 ... [ ok ] * Bringing up interface enp5s0 * Changing MAC address of enp5s0 ... [ ok ] * changed to xx:xx:xx:xx:xx:xx * xxxx:xxxx:xxxx:xxxx::2/64 ... [ ok ] * 192.168.178.2/24 ... [ ok ] * Adding routes * default via 192.168.178.1 ... [ ok ] * default via fe80::0 ... [ ok ] * Activating additional swap space ... [ ok ] * setting up tmpfiles.d entries ... [ ok ] * Initializing random number generator ... [ ok ] rc boot logging stopped at Sat Jun 22 01:11:42 2013
> * Configuring kernel parameters ... > [ ok ] means that sysctl service have runned, it should apply sysctl -p to every readable file. 1). can you run sysctl -p /etc/sysctl.conf or /etc/sysctl/local.conf to check if values are applied by sysctl. If they are applied 2).can you run: /etc/init.d/sysctl zap # clean up real sysctl state /etc/init.d/sysctl -d start 2&> log # start sysctl once again and paste file log here. Thanks.
Created attachment 351674 [details] log
(In reply to Darko Luketic from comment #7) > Created attachment 351674 [details] > log Thanks. Strange according to logs all conf files are applied without errors. ++ sysctl -p /etc/sysctl.conf ... ++ sysctl -p /etc/sysctl.d/local.conf It seems for me that some other script overrides this settings. It will need some more investigation.
thank you too. bootmisc | boot fsck | boot hostname | boot hwclock | boot keymaps | boot localmount | boot lvm | boot mcelog | boot modules | boot mtab | boot net.lo | boot procfs | boot root | boot swap | boot swapfiles | boot sysctl | boot termencoding | boot tmpfiles.setup | boot urandom | boot zfs | boot alsasound | default dbus | default distccd | default lm_sensors | default local | default metalog | default net.enp6s0 | default netmount | default nfs | default nginx | default ntpd | default preload | default smartd | default sshd | default uwsgi | default vixie-cron | default xdm | default killprocs | shutdown mount-ro | shutdown savecache | shutdown devfs | sysinit dmesg | sysinit sysfs | sysinit udev | sysinit udev-mount | sysinit I've also searched Google and it seems like others also have this problem, at least I saw some posts on the forum without a real solution, only workarounds. /etc/local.d/ is empty except for the README. Maybe it has something to do with sysfs my fstab, do I need to mount /proc and so on in there? /dev/sdc2 /boot ext2 noauto,noatime 0 2 /dev/sda6 / ext4 defaults,noatime 0 1 /dev/sdc1 /boot/efi vfat noauto,noatime 0 0 /dev/sdc3 none swap sw 0 0 tosh/var /var zfs defaults 0 0 /dev/cdrom /mnt/cdrom auto noauto,user 0 0 tmpfs /tmp tmpfs nosuid 0 0 vartmpfs /var/tmp/portage tmpfs uid=250,gid=250,mode=0775 0 0 #server:/storage /home/darko/srv nfs _netdev,defaults,user,auto,noatime,intr,rw 0 0 192.168.178.2:/usr/portage /usr/portage nfs _netdev,defaults,auto,noatime,intr,rw 0 0 192.168.178.2:/usr/local/portage /usr/local/portage nfs _netdev,defaults,auto,noatime,intr,rw 0 0
Currently I don't see a reason for such behaviour. I'll try to configure my system on the week to check your problem more precisely, (as I have use_tempaddr=2 as a default value, and I have no real ipv6 interfaces to test). Currently you can try to make next steps: 1). add debug logging to sysctl script itself, i.e. smth like: in the end of the start() function in /etc/init,d/sysctl echo "after sysctl" >> /var/log/findsysctlbug cat /proc/sys/net/ipv6/conf/all/use_tempaddr >> /var/log/findsysctlbug and check whether the value was really set, if no then we need to search problem in sysctl itself, or in load order. 2). add you sysctl service to local one: echo "sysctl -p /etc/sysctl.conf" >> /etc/local.d/mysysctl.start and possibly add debug script. This will gave an answer if problem may be related to load order. Sorry, that I'm not able to do in myself now, I'll try to do it as soon as it will be possible. And maybe other part of our team have more ideas how to track this problem.
ok... here goes... it's not related to ipv6 only the ipv6 problem with sysctl is solved by adding the ipv6 module to /etc/conf.d/modules co-incidently distcc refused to start because DISTCCD_OPTS="${DISTCCD_OPTS} --listen 192.168.178.20" was set although an ipv4 lease was acquired (but I guess the iface wasn't up so fast) removing the config option "solved" this as it's now listening on 0.0.0.0, which might or might not be the desired behaviour. vm.swappiness=10 vm.dirty_background_ratio=50 vm.dirty_ratio=80 is still affected sysctl vars are set with the init script 80 but are reverted to some default value 10 however only on the desktop. it works on the server differences are, a different kernel, different hardware, server does not have cinnamon running or xdm or lightdm or preload or a pre-emptive kernel, has a static ipv4 and ipv6. I didn't try 2.) yet
correction vm.swappiness is set but vm.dirty_background_ratio=50 vm.dirty_ratio=80 aren't
I've tested on a fresh system and vm.swappiness, vm.dirty_background_ratio, vm.dirty_ratio worked for me. If you want I can try to prepare additional log scripts in order to track changes in verify that options are really set, and track when they are changed.
ok, but I can't promise a timely response, lots of stuff to do right now I'll squeeze it in where I can. So go ahead :)
I suspect the problem is that sysctl can't load/modify parameters for modules that aren't loaded yet (and/or they'll get stomped on when the module does load). I'm looking at the same thing now with conntrack tweaks. This seems to be a common problem. The Gentoo sysctl(8) man page doesn't mention it, but see for example this version of sysctl.8: http://linux.die.net/man/8/sysctl And some bugs for other distros: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=507788 https://bugs.launchpad.net/ubuntu/+source/procps/+bug/50093 The workarounds I've seen proposed include: - Make sure all modules whose sysctl's you plan to tweak are already loaded; in our case that means explicitly listing them in /etc/conf.d/modules - put off running the sysctl init.d service later (possible security downsides) - load sysctl.conf again later, in /etc/local.d/ or similar (ugly) - arrange for modules to get autoloaded if their /proc/sys files don't exist (ugly, complex, and ugly)
(In reply to Hank Leininger from comment #15) > I suspect the problem is that sysctl can't load/modify parameters for > modules that aren't loaded yet (and/or they'll get stomped on when the > module does load). I'm looking at the same thing now with conntrack tweaks. > > This seems to be a common problem. The Gentoo sysctl(8) man page doesn't > mention it, but see for example this version of sysctl.8: > > http://linux.die.net/man/8/sysctl > > And some bugs for other distros: > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=507788 > > https://bugs.launchpad.net/ubuntu/+source/procps/+bug/50093 > > The workarounds I've seen proposed include: > > - Make sure all modules whose sysctl's you plan to tweak are already loaded; > in our case that means explicitly listing them in /etc/conf.d/modules This is the cleanest way to handle this, so it isn't a bug in OpenRC, just make sure that you list all of the modules you need in /etc/conf.d/modules or build these modules into the kernel.