**** BEGIN LOGGING AT Wed Nov 09 02:59:57 2011 Nov 09 04:57:03 build #98 of ramips is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/ramips/builds/98 Nov 09 06:58:00 hm, I wonder why menuconfig segfaults for me when searching Nov 09 07:15:42 {Nico}: At the moment btrfs doesn't show up in menuconfig unless the dependencies are selected manually. Maybe that could be reversed? Nov 09 07:16:18 {Nico}: also, kmod-lib-crc32c should be moved to the lib-submenu (not crypto)? Nov 09 07:34:52 loswillios * r28873 /packages/net/weechat/Makefile: [packages] weechat: update to 0.3.6 Nov 09 08:29:19 loswillios: try adding + infront of dependencies to get what you want Nov 09 08:29:27 for brtfs Nov 09 09:17:03 build #98 of pxcab is complete: Failure [failed compile_7] Build details are at http://buildbot.openwrt.org:8010/builders/pxcab/builds/98 Nov 09 09:19:05 uhm, im not sure if you guys have the time for this but it would be cool if you can break the iptables nat extra into pieces for tftp, ftp, irc, gre, pptp so we can compile only what we need Nov 09 10:08:49 and btrfs needs two patches to work: http://pastebin.com/raw.php?i=0ZAmd0e4 and http://patchwork.openwrt.org/patch/1199/ Nov 09 10:10:00 build #106 of at91 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/106 Nov 09 10:26:30 build #104 of ubicom32 is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/ubicom32/builds/104 Nov 09 10:30:46 typo: http://pastebin.com/raw.php?i=sLbzEza6 Nov 09 10:59:41 build #95 of rb532 is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/95 Nov 09 11:11:14 jow * r28874 /trunk/package/firewall/ (Makefile files/firewall.config): [package] firewall: add DHCPv6 default rule (#10381) Nov 09 11:13:19 jow * r28875 /branches/backfire/package/firewall/ (Makefile files/firewall.config): [backfire] firewall: merge r28874 Nov 09 11:16:03 thanks jow ! Nov 09 11:18:47 nico * r28876 /branches/packages_10.03.1/libs/neon/Makefile: [backfire] packages/neon: merge r28842 Nov 09 11:20:54 nico * r28877 /branches/packages_10.03.1/libs/openldap/ (Makefile files/ldap.init): [backfire] packages/openldap: merge r28843 Nov 09 11:25:44 nico * r28878 /branches/packages_10.03.1/net/apache/Makefile: [backfire] packages/apache: merge r28845, r28846 & r28847 Nov 09 11:48:22 juhosg * r28879 /trunk/package/iw/ (Makefile patches/120-per_chain_signal_strength.patch): (log message trimmed) Nov 09 11:48:22 pacakge/iw: don't show a stray ']' if the chain signal attribute list is empty Nov 09 11:48:22 Without the patch, the output of 'station dump' command looks weird in Nov 09 11:48:22 some cases: Nov 09 11:48:22 root@OpenWrt:/# iw dev wlan0 station dump Nov 09 11:48:22 Station 00:21:27:cb:57:02 (on wlan0) Nov 09 11:48:23 inactive time: 190 ms Nov 09 11:51:40 ping {Nico} Nov 09 11:51:42 hi Nov 09 11:51:51 <{Nico}> tripolar: hi & pong Nov 09 11:52:33 {Nico}: how can i kill all running instanced of a service with the new service wrapper? ssh for example hangs when i reboot and i dont kill all instances Nov 09 11:54:06 instanced/instances Nov 09 11:54:20 <{Nico}> service_stop COMMAND will stop all process matching COMMAND Nov 09 11:54:49 <{Nico}> if SERVICE_USE_PID or SERVICE_PID_FILE are set, it will alse match the pid in the pid-file Nov 09 11:55:34 but if i set SERVICE_USE_PID=1 in the beginning of the init script it doesnt - it just kills the instance thats pid is in /var/run/sshd.pid Nov 09 11:55:59 <{Nico}> so if you want to kill all instances regarding the pid, you can try this 'SERVICE_USE_PID= service_stop /usr/sbin/sshd' fr example Nov 09 12:03:28 build #92 of ppc44x is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/92 Nov 09 12:04:18 {Nico}: this doesnt work Nov 09 12:04:58 <{Nico}> tripolar: show me your initscript Nov 09 12:07:42 {Nico}: http://nopaste.dk/p4206 Nov 09 12:08:25 also starting and stopping lit like this doesnt work SERVICE_USE_PID=1 service_start /usr/sbin/sshd Nov 09 12:09:27 the process also looks like this {sshd} sshd: root@pts/0 - and the service looks like this /usr/sbin/sshd in ps aux output Nov 09 12:09:47 {sshd} sshd: root@pts/0 == the root login Nov 09 12:12:52 shutdown() {killall sshd} ; works Nov 09 12:13:42 killall /usr/sbin/sshd doesnt Nov 09 12:17:14 <{Nico}> try running your initscript like this 'SERVICE_QUIET= /etc/init.d/sshd ACTION' Nov 09 12:17:55 but then stop and shutdown would look the same Nov 09 12:18:57 <{Nico}> it's just for debug (it will show the output of start-stop-daemon) Nov 09 12:21:33 i just get the output stopped /usr/sbin/sshd (pid 2692) Nov 09 12:22:54 shutdown now looks like - SERVICE_USE_PID= SERVICE_QUIET= service_stop /usr/sbin/sshd Nov 09 12:25:25 and it doesnt work Nov 09 12:26:34 {Nico}: should i stay with killall in shutdown? Nov 09 12:26:56 killall is always bad in initscripts Nov 09 12:27:22 <{Nico}> can you show me the result of 'ps ax | grep ssh' Nov 09 12:27:41 murphy says killall will always kill the script starting it first if it has the same name Nov 09 12:28:01 2656 root 5420 S {sshd} sshd: root@pts/0 Nov 09 12:28:01 2715 root 2804 S /usr/sbin/sshd Nov 09 12:28:01 2733 root 1448 S grep ssh Nov 09 12:28:41 <{Nico}> did you try SERVICE_USE_PID=1 service_stop {sshd} ? Nov 09 12:28:51 no Nov 09 12:30:09 no {sshd} found; none killed Nov 09 12:30:13 is the debug output Nov 09 12:30:23 the same for sshd only Nov 09 12:30:30 <{Nico}> cat /proc/2656/cmdline please Nov 09 12:30:48 root=/dev/mtdblock2 rootfstype=squashfs,jffs2 noinitrd Nov 09 12:31:28 <{Nico}> not /proc/cmdline Nov 09 12:31:42 sshd: root@pts/0 Nov 09 12:31:48 yeah wrong file :) Nov 09 12:32:10 <{Nico}> then try SERVICE_USE_PID=1 service_stop sshd: Nov 09 12:32:21 <{Nico}> then try 'SERVICE_USE_PID= service_stop sshd:', sorry Nov 09 12:32:42 no sshd: found; none killed Nov 09 12:33:50 <{Nico}> cat /proc/2656/stat please Nov 09 12:34:04 2656 (sshd) S 1 2656 2656 0 -1 4194560 568 1272 0 0 392 376 246 14 20 0 1 0 8666 5550080 370 2147483647 4194304 4538468 2139353856 2139351992 2006390752 0 0 4096 147462 14999983 0 0 18 0 0 0 0 0 0 Nov 09 12:37:07 <{Nico}> ok, i have to change the service function to allow matching processes by name as well Nov 09 12:57:33 {Nico}: if you have a finished script i can test it ... Nov 09 13:21:25 <{Nico}> tripolar: it turns out if i match by process names, i will also kill the initscript, which is exactly what killall is doing :) Nov 09 16:11:45 tripolar * r28880 /packages/net/nfs-kernel-server/files/nfsd.init: [packages] nfs-kernel-server: use new service wrapper Nov 09 16:40:56 tripolar * r28881 /packages/net/ntpd/files/ntpd.init: [packages] ntpd: use new service wrapper Nov 09 16:50:48 tripolar * r28882 /packages/net/ntpd/files/ntpd.init: [packages] ntpd: let ntpd create the pid file Nov 09 16:52:18 tripolar * r28883 /packages/net/ntpd/files/ntpd.init: [packages] ntpd: remove uneeded line Nov 09 17:03:34 tripolar * r28884 /packages/net/nfs-kernel-server/files/nfsd.init: [packages] nfsd: only demonize and write pid for rpc.mountd Nov 09 17:33:02 nico * r28885 /trunk/package/busybox/ (2 files in 2 dirs): package/busybox: passwd: use MD5 hash by default (like it used to be) Nov 09 17:35:08 yuck... Nov 09 18:09:39 oh lol Nov 09 18:10:12 something must go wrong during sysupgrade, I am still at r28709 according to /rom/etc/banner Nov 09 18:18:02 can someone please look at patchworks 1579-1581? these are upstreamed into netdev. what other platforms use the solos chipset? I thought one of the AR71xx boxes also used it. Nov 09 18:26:22 anyone know if backfire-rc6 supports ethernet bridging on 'regular' (and tun/tap) interfaces? Nov 09 18:49:31 On trunk the following works, but not on backfire: On the VPN 'server': openvpn and a bunch of quicktun tunnels. The tap devices of openvpn and quicktun are bridged (and appear correctly in brctl show). On one of the remote quicktun nodes I 'nslookup hostname qtvpnserver') Nov 09 18:50:02 when I tcpdump on the server it works on backfire, (it always works on trunk), but not without tcpdump Nov 09 18:50:33 I'm led to believe that this is a result of promisc mode on for the tcpdump, but not otherwise Nov 09 18:56:08 jow * r28886 /trunk/package/uhttpd/ (Makefile src/uhttpd-utils.c): [package] uhttpd: cope with DES crypted passwd entries by not relying on a leading dollar sign to indicate a cipher Nov 09 18:57:19 stintel: I added some code to sysupgrade which kills a bunch of processes Nov 09 18:57:21 jow_laptop: hmmm. Will we still be able to use mkpasswd -m md5 (for example) on debian to generate passwords or is passwd now busybox specific Nov 09 18:58:03 stintel: see whether it maybe kills udhcpc so that you loose the connection Nov 09 18:58:23 cshore: no will still work, uhttpd just was coded with the wrong assumption that any kind of hash starts with $ Nov 09 18:58:43 but actually $x$... is a glibc extension Nov 09 18:58:54 and only applicable to non-DES ciphers Nov 09 18:59:51 which the lead to the discovery that bb changed its default crypt cipher due to some patch Nov 09 18:59:58 nbd: Thanks for the heads up on the git-peek-remote and I appreciated that. Nov 09 19:00:13 well manually turning on promisc works, so perhaps some vpn thing is turning it off on me Nov 09 19:00:39 I thought the bridge forces all its ports to be in promisc mode Nov 09 19:00:47 or do you refer to the bridge itself? Nov 09 19:00:56 br-qtlan (for instance) Nov 09 19:01:46 jow_laptop: Re: passwd: ah ok, thanks for that Nov 09 19:02:13 hmmm...apparently the problem could have been using tcdump without -p Nov 09 19:02:20 *tcpdump Nov 09 19:02:27 let me reboot and see Nov 09 19:08:53 no such luck Nov 09 19:12:12 is the br-qtlan supposed to need promisc? I think maybe there is some confusion on that regard...I hadn't heard of that before. If not then I've got to figure out what promisc on at that level is doing that makes it work. Nov 09 19:13:08 it's not the firewall...I've had logging on on both sides and there are no dropped packets, and I've looked at the iptables output and it's got the right rules Nov 09 19:13:53 ssh and syslog work, so the tunnel is fine Nov 09 19:14:01 it's driving me batty Nov 09 19:14:59 I don't know what DNS does that's different....time to look up the RFC I guess Nov 09 19:17:26 oh wait those are between bridged interfaces, no to the vpn router Nov 09 19:18:13 DNS is to the VPN router, and ping isn't working either...but no firewall rejects, and promisc 'fixes' it Nov 09 19:21:12 hmm...tested from another quicktun node and accessing DNS to another bridge interface works, but not to the router Nov 09 19:21:21 *bridged Nov 09 19:23:25 ok, SSH to the router when to the bridge interface, so it looks like it's all 'to the router' traffic that isn't showing up. Nov 09 19:23:45 *SSH to the router fails Nov 09 19:23:57 but only on the vpn bridge Nov 09 19:24:00 br-lan is fine Nov 09 19:25:48 I have a sneaking suspicion the problem is the older openvpn Nov 09 19:34:53 nope, this time it wasn't the problem (I had an issue with openvpn a long time ago which is why I thought it might be that) Nov 09 19:38:05 so the question is why there is no traffic to the router through the bridge, but between tunnels (through the bridge) is fine. I'm getting the feeling I'm forgetting some basic networking thing that will be completely obvious afterwards. Nov 09 19:40:27 dropping the firewall on incoming VPN connections to see if despite no log messages it's the firewall Nov 09 19:41:14 nope Nov 09 19:48:46 hmmm....iptables counters are 0 for the accept rules that would apply, so probably the firewall isn't getting the traffic. Why would a br-qtlan not get traffic when it's components are able to talk to each other? Nov 09 19:49:35 nico * r28887 /packages/admin/monit/ (Makefile files/monit.init): packages/monit: use new service functions, fix /etc/monitrc permissions Nov 09 19:49:38 nico * r28888 /packages/admin/nrpe/ (Makefile files/nrpe.init): packages/nrpe: use new service functions, move user/group creation from postinst to initscript Nov 09 19:49:39 nico * r28889 /packages/admin/osiris/ (Makefile files/osirisd.init): packages/osiris: use new service functions, set /var/lib/osiris perms Nov 09 19:49:41 nico * r28890 /packages/admin/pmacct/ (5 files in 2 dirs): packages/pmacct: use new service functions Nov 09 19:49:44 nico * r28891 /packages/admin/ (4 files in 4 dirs): packages/syslog-ng{,3}: use new service functions Nov 09 19:49:46 nico * r28892 /packages/admin/zabbix/ (Makefile files/zabbix_agentd.conf files/zabbix_agentd.init): Nov 09 19:49:47 packages/zabbix: use new service functions, various other zabbix-agent fixes: Nov 09 19:49:47 * move user/group creation from postinst to initscript Nov 09 19:49:47 * use /var/run/zabbix directory for pid file Nov 09 19:49:47 * use /var/log/zabbix directory for log file Nov 09 19:49:47 * set correct ownership and perms for both directories Nov 09 19:55:53 if ARCH_REQUIRE_GPIOLIB is selected, should GPIO_SYSFS be included by default? Nov 09 19:56:13 or is it right/normal that you have to do a make kernel_menuconfig to enable it yourself? Nov 09 19:56:40 or, should I select it by default for a platform? Nov 09 20:05:13 jow_laptop: explains why the sysupgrade always finished so fast lately ;) Nov 09 20:07:18 jow_laptop: device has a static IP though. and seems sysupgrade via command line also immediately disconnects me Nov 09 20:07:44 stintel: do you screen or something? Nov 09 20:08:26 stintel: or some other wrapper around the shell? (e.g. a vpn or something other than dropbear directly in?) Nov 09 20:08:38 cshore: screen will also be killed ;) Nov 09 20:08:49 I should probably just edit the killall line Nov 09 20:08:50 stintel: yes, that's what i was saying Nov 09 20:09:18 cshore: ah, misunderstood :) I tried in screen after your question Nov 09 20:09:57 stintel: try thing sh -c "( sysupgrade -n image.trx & ) &" & Nov 09 20:10:03 that will detach it from the terminal Nov 09 20:10:13 more than you need actually but it makes sure Nov 09 20:12:29 I commented the kill line in /lib/upgrade/common.sh Nov 09 20:18:43 stintel: tell me if you still have the issue with a recent build later, ideally I want it to work reliably ;) Nov 09 20:19:50 jow_laptop: will do Nov 09 20:20:04 ah sweet, and now I also have the rngd Nov 09 20:20:13 lol, should have checked /etc/banner earlier Nov 09 20:23:07 flashing main router now, hope it comes back :) Nov 09 20:31:43 ok that worked. seems netif doesn't support 6in4 yet? Nov 09 20:32:40 right Nov 09 20:47:10 build #107 of s3c24xx is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/107 Nov 09 21:12:49 ok apparently in backfire a bridge has to have a physical interface or it doesn't quite work Nov 09 21:17:18 jogo * r28893 /trunk/ (14 files in 11 dirs): Nov 09 21:17:19 target: linux: mtd: fix MTDREFRESH to an arbitrary high number Nov 09 21:17:19 To make the ioctl number "stable", use an arbitrary high number to prevent Nov 09 21:17:19 conflicts with new mtd ioctls that would push MTDREFRESH's number. Nov 09 21:17:19 Also make mtd use the in-kernel mtd headers. Nov 09 21:17:22 jogo * r28894 /trunk/package/madwifi/Makefile: package: madwifi: mark as broken for 3.2 Nov 09 21:17:25 jogo * r28895 /trunk/package/acx/Makefile: package: acx: mark as broken for 3.2 Nov 09 21:17:28 jogo * r28896 /trunk/package/mac80211/Makefile: package: mac80211: mark as broken for 3.2 Nov 09 21:17:32 jogo * r28897 /trunk/package/kernel/modules/ (crypto.mk netdevices.mk netsupport.mk other.mk spi.mk): package: kernel: add handling for 3.2 renames (and some missed ones for 3.1) Nov 09 21:38:38 build #104 of brcm47xx is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx/builds/104 Nov 09 21:58:31 build #90 of uml is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/90 Nov 09 22:12:42 jogo * r28898 /trunk/package/kernel/modules/crypto.mk: package: kernel: fix typo Nov 09 23:12:24 nico * r28899 /packages/net/openssh/ (Makefile files/sshd.init): packages/openssh: use new service functions, move user/group creation from postinst to initscript Nov 09 23:18:01 nico * r28900 /packages/net/leafnode/ (Makefile files/leafnode.init): packages/leafnode: move user/group creation from postinst to an initscript Nov 09 23:18:04 nico * r28901 /packages/net/prosody/ (Makefile files/prosody.init): packages/prosody: move user/group creation from postinst to initscript Nov 09 23:18:06 nico * r28902 /packages/net/ (6 files in 6 dirs): packages/tor{,-alpha}: use new service functions, move user/group creation from postinst to initscript, move pid file to /var/run/tor directory Nov 09 23:25:56 nico * r28903 /packages/mail/dovecot/ (Makefile files/dovecot.init): packages/dovecot: use new service functions, move user/group creation from postinst to initscript Nov 09 23:25:58 nico * r28904 /packages/libs/howl/ (4 files in 2 dirs): packages/howl: use new service functions Nov 09 23:25:59 nico * r28905 /packages/libs/openldap/ (Makefile files/ldap.init): packages/openldap: use new service functions Nov 09 23:30:09 nico * r28906 /packages/net/rsync/ (Makefile files/rsyncd.init): packages/rsync: use new service functions Nov 09 23:30:10 nico * r28907 /packages/net/vsftpd/ (Makefile files/vsftpd.init): packages/vsftpd: use new service functions Nov 09 23:30:13 nico * r28908 /packages/net/xinetd/ (Makefile files/xinetd.init): packages/xinetd: use new service functions Nov 09 23:37:42 build #85 of ixp4xx is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/ixp4xx/builds/85 Nov 09 23:38:27 nbd: ping Nov 09 23:38:31 luka12345|wiik: pong Nov 09 23:38:37 nico * r28909 /packages/utils/at/ (Makefile files/atd.init): packages/atd: use new service functions Nov 09 23:38:39 nico * r28910 /packages/net/cups/ (Makefile files/cupsd.init): packages/cups: use new service functions Nov 09 23:38:41 nico * r28911 /packages/net/ipsec-tools/ (Makefile files/racoon.init): packages/ipsec-tools: use new service functions Nov 09 23:38:47 hi, great that you are here Nov 09 23:39:43 i have part of my config file like this: http://pastebin.com/QKqWapza Nov 09 23:39:52 but i have only one device Nov 09 23:39:55 no more Nov 09 23:40:16 so do i need to use libubox Nov 09 23:40:25 for what? Nov 09 23:40:25 or i can just use uci Nov 09 23:40:34 you can use uci Nov 09 23:40:38 to get the data to my struct Nov 09 23:40:53 depends on what you want to do Nov 09 23:41:12 as i see in the code libubox is nice when you have for exampe 10 devices Nov 09 23:41:29 and you want to loop over them or something right? Nov 09 23:41:33 but i dont need that Nov 09 23:41:40 i have only one device Nov 09 23:42:05 yeah, it's definitely for more complex scenarios Nov 09 23:43:45 so there is no need for libubox to copy the data from the config file from above to struct like this: Nov 09 23:43:49 http://pastebin.com/GaDFZtZB Nov 09 23:44:05 that will make my life much easier Nov 09 23:44:06 nico * r28912 /packages/net/arpwatch/ (Makefile files/arpwatch.init): packages/arpwatch: use new service functions Nov 09 23:44:07 nico * r28913 /packages/net/oidentd/ (Makefile files/oidentd.init): packages/oidentd: use new service functions Nov 09 23:44:13 nico * r28914 /packages/net/opennhrp/ (Makefile files/opennhrp.init): packages/opennhrp: use new service functions Nov 09 23:44:19 nico * r28915 /packages/net/parprouted/ (Makefile files/parprouted.init): packages/parprouted: use new service functions Nov 09 23:44:20 nico * r28916 /packages/net/ptpd/ (Makefile files/ptpd.init): packages/ptpd: use new service functions Nov 09 23:44:27 nico * r28917 /packages/utils/rng-tools/ (Makefile files/rngd.init): packages/rng-tools: use new service functions Nov 09 23:44:28 nico * r28918 /packages/utils/smartmontools/ (Makefile files/smartd.init): packages/smartmontools: use new service functions Nov 09 23:44:30 nico * r28919 /packages/net/ulogd/ (Makefile files/ulogd.init): packages/ulogd: use new service functions Nov 09 23:44:31 nico * r28920 /packages/net/keepalived/ (Makefile files/keepalived.init): packages/keepalived: use new service functions Nov 09 23:47:47 nico * r28921 /packages/net/rarpd/ (Makefile files/rarpd.init): packages/rarpd: use new service functions Nov 09 23:47:52 nbd: in device.h lines 34 - 40 and device.c lines 113 - 120 Nov 09 23:48:36 can you tell me how is that type of programing in c called, i would like to google for some tutorial Nov 09 23:49:54 nico * r28922 /packages/ipv6/ahcpd/ (Makefile files/ahcpd.init): packages/ahcpd: use new service functions Nov 09 23:49:56 nico * r28923 /packages/ipv6/gw6c/ (Makefile files/gw6c.init): packages/gw6c: use new service functions Nov 09 23:49:58 nico * r28924 /packages/ipv6/miredo/ (Makefile files/miredo-server.init files/miredo.init): packages/miredo: use new service functions Nov 09 23:50:00 nico * r28925 /packages/ipv6/mrd6/ (Makefile files/mrd6.init): packages/mrd6: use new service functions, add a dependency on kmod-ipv6 Nov 09 23:50:04 nico * r28926 /packages/ipv6/ptrtd/ (Makefile files/ptrtd.init): packages/ptrtd: use new service functions Nov 09 23:50:05 nico * r28927 /packages/ipv6/radvd/ (Makefile files/radvd.init): packages/radvd: use new service functions Nov 09 23:50:07 nico * r28928 /packages/ipv6/totd/ (Makefile files/totd.init): packages/totd: use new service functions Nov 09 23:50:08 nico * r28929 /packages/ipv6/wide-dhcpv6/ (Makefile files/dhcp6s.init): packages/wide-dhcpv6: use new service functions Nov 09 23:50:11 nico * r28930 /packages/utils/cmdpad/ (Makefile files/cmdpad.init): packages/cmdpad: use new service wrapper Nov 09 23:50:13 luka12345|wiik: no idea what that type of programming is called Nov 09 23:50:21 luka12345|wiik: i came up with it after many iterations and experiments Nov 09 23:51:30 {Nico}: hi, why did you create this dir in the openssh init script? Nov 09 23:51:33 /var/run/sshd Nov 09 23:51:40 ok... i'm going to write a nice and simple config.c now ;) Nov 09 23:51:44 sshd just uses /var/run/sshd.pid Nov 10 00:11:06 tripolar * r28931 /packages/net/openssh/files/sshd.init: Nov 10 00:11:07 [packages] openssh: * set ssh dir to /var/empty Nov 10 00:11:07 * group_add sshd not ssh Nov 10 00:11:07 * dont create /var/run/sshd Nov 10 00:12:32 <{Nico}> tripolar: it is sshd's homedir, created it just to be consistent Nov 10 00:15:36 {Nico}: the homedir was defined as /var/empty/.sshd Nov 10 00:15:47 i defined it to /var/empty Nov 10 00:16:22 as sshd uses this dir befor it drops it root rights (when you login as use != root) Nov 10 00:16:32 use/user Nov 10 00:18:01 gentoo even uses /dev/null Nov 10 00:18:37 <{Nico}> yeah, i know that, but thought the purpose of /var/empty was to remain empty, not to be someone's homedir Nov 10 00:19:12 ssh never uses this dir as homedir - its not possible to login as ssh user directly Nov 10 00:20:24 <{Nico}> yeah, i know that toot Nov 10 00:21:10 <{Nico}> i was just lazy and just thought: if i create a user, and it's homedir is /var/run/sshd, then i shoul create /var/run/sshd, that's all Nov 10 00:21:20 http://www.openssh.org/openbsd.html - openssh even explains it here Nov 10 00:21:39 it was wrong befor too but now it should be right Nov 10 00:22:58 <{Nico}> no worries Nov 10 00:23:12 <{Nico}> is the shutdown function working for you? Nov 10 00:23:26 yes it works fine - thanks Nov 10 00:23:41 <{Nico}> np Nov 10 00:24:28 <{Nico}> i tried putting that in the service_* wrappers, but it became to ugly :) Nov 10 00:28:34 {Nico}: okay have a nice night i'll go to sleep Nov 10 00:33:50 <{Nico}> tripolar: 'night **** ENDING LOGGING AT Thu Nov 10 02:59:58 2011