**** BEGIN LOGGING AT Fri Jul 22 23:59:56 2005 Jul 23 02:35:38 DaKa2 ping Jul 23 03:49:20 I have to fix my irssi... Jul 23 06:17:12 Downloaded the monotone snapshort of 2.3beta, built and installed. It boots nicely and things seem ok (any NFS problems would go unnoticed here anyway). The ipkg feeds for it are not yet operational I see. Should I disregard and just change the feed URL? Jul 23 06:28:17 feeds updated Jul 23 06:46:08 ncftp works it seems. Straight forward compile. I have a problem with syslog-ng - it conflicts with syslog in the firmware, but there is no package for it I guess. Should I manually remove it on install of syslog-ng? Jul 23 06:51:59 It is the same with the busybox conflicts. Wouldn't it be nice it e.g. "grep" removed the busybox link "grep" (asking first) so it would acitvate itself ? At least it should warn that another grep existed? Jul 23 06:52:52 ingeba-toandfrom: a big fix from DaKa2 which has just been committed does just that (update-alternatives for busybox) Jul 23 06:53:11 * \o/ * Jul 23 07:16:29 xinetd runs nicely - thanks DaKa2 ! proftpd seems to be ok using it. I will not support other inetd's in the proftpd package Jul 23 07:16:45 :-) Jul 23 07:17:11 * ingeba-toandfrom starts to believe openslug will be a good friend. Jul 23 10:23:00 Anyone knows where/if libstdc++ can be found on Openslug? Jul 23 10:28:06 ipkg install libstdc++6 ? Jul 23 10:31:35 I'm blind - only looked in packages. Jul 23 10:31:50 and it is not there Jul 23 10:33:04 I can RDEPEND on it? Jul 23 10:33:34 I guess so.. Jul 23 10:33:50 I'll try. Jul 23 10:36:11 DaKa2: Yes - it worked Jul 23 10:36:31 :) Jul 23 11:03:01 man pages working :-) Jul 23 11:04:01 dammit, that means I have to update some of my packages to actually packages them.. Jul 23 11:04:08 :-D Jul 23 11:04:54 I love OE! Jul 23 11:05:59 :) Jul 23 11:06:36 does that mean there is one unslung developer less now? Jul 23 11:07:19 DaKa2: Soon it will be, except for modules. I will still maintain, though. Basically the same packages anyhow Jul 23 11:07:56 I'm pretty happy with unslung except for the USB limitations. Jul 23 11:08:10 :) Jul 23 11:09:09 It is dead stable in ftp/samba/ssh environment - have had my slug running for months until I had to upgrade... Jul 23 11:09:59 with a little work openslug can be that stable too, its just hard to not play with it .-) Jul 23 11:10:25 DaKa2: Are all pkgs supposed to go in oe-symlinks or will we split into basic oe and opt oe? Jul 23 11:10:49 I have new 5 packages in testing now Jul 23 11:11:04 everything for openslug should be in oe-symlinks Jul 23 11:11:12 I think :-) Jul 23 11:11:23 Ok - I will cry for someone to hold my hand when commit-time comes. Jul 23 11:11:49 Do we have a testing area? Oh, I'll as NAiL Jul 23 11:13:11 What now? :) Jul 23 11:18:17 NAiL ! I was wondering if we have a ready-for-testing state for new packages - don't trust my coding - experience tells me... Jul 23 11:18:44 Also I had to change several outdated OE pkgs Jul 23 11:19:03 to use newer versions. Man was a total mess as well. Jul 23 11:19:45 So I do not dare committing any of them. "They" will hunt me down and come for me with pitch forks Jul 23 11:51:06 There is no testing currently, other than people giving feedback from the unstable feed Jul 23 11:51:35 ie, all packages that you want tested go into unstable (provided they *build*) Jul 23 11:53:03 ingeba-toandfrom: what packages have you updated? Jul 23 11:53:59 NAiL: Will you hold my hand when I commit. Prob. tomorrow - want to get things tested a bit here. Was very delayed getting en recent OpenSlug version compiled and have to give back the dev slug early next week. Jul 23 11:54:38 ingeba-toandfrom: sure. Would you mind mailing me the diff so I can look at it? Jul 23 11:55:05 I'll mail you a tarball of all new bb files and diffs for the changed ones. ok? Jul 23 11:55:11 thanks Jul 23 11:55:17 as well as auxillary files. Jul 23 11:55:48 good Jul 23 11:56:52 NAiL: Can you advice me on syslog-ng. It needs to disable the fw syslog so not "RCONFLICT" can solve it. Should I offer to remove the init script? Jul 23 11:57:24 hmm... good question Jul 23 11:57:29 Or just detect and ask them to do it manually Jul 23 11:58:06 detect&manual is not always a good idea, since "ipkg upgrade" might scroll that part by Jul 23 11:58:22 Dunno if it's possible to do interactive stuff in pkg_postinst Jul 23 11:58:33 ouch Jul 23 11:58:37 no... not a good idea Jul 23 11:58:49 There are graphical frontends to ipkg Jul 23 11:59:03 Maybe not, so I can at least detect and exit before installing. Saying "You must do this to unstall syslog and this to install syslog-ng" Jul 23 11:59:05 which probably wouldn't understand that bit Jul 23 11:59:12 Oh dear Jul 23 11:59:26 doing the detect stuff is the best idea imho Jul 23 11:59:36 And fail if present? Jul 23 12:00:13 Yeah. Fail and print a helpful message how to fix Jul 23 12:00:29 preferrably with "copy&paste"-able cmds Jul 23 12:00:35 Will do. Jul 23 12:01:10 thanks :) Jul 23 12:01:43 Upgraded packages are ncftp (symlink added) and man (which is also patched - it seemed totally broken). Jul 23 12:01:53 Added proftpd, libol, syslog-ng and groff Jul 23 12:02:20 symlinks added for man and man-pages Jul 23 12:03:17 svn is down, so I can't add the symlinks yet Jul 23 12:03:42 Dont worry - wont do a thing until tomorrow and maybe the Germans have put their act together by then. Jul 23 12:04:10 someone said that berlios wasn't monitored during weekends, so it might not get up until monday Jul 23 12:04:27 but this might be a good time to look at alternate ways of implementing the symlinks anyway Jul 23 12:04:30 Can we set up mirrors in the long run? Jul 23 12:04:43 svn mirror you mean? Jul 23 12:04:47 Yes Jul 23 12:05:13 For bitbake and the other stuff - like it is today noone can set up a dev system Jul 23 12:05:14 There was talk about moving the svn stuff to nudi (ie, the official dev box where monotone lives) Jul 23 12:05:42 Ok - mirrors are often good anyway, but I know too little of the implications Jul 23 12:06:09 I don't know enough about svn to say if it's easy/possible to do mirroring Jul 23 12:08:40 Anyone got any good suggestions on alternatives to svn for the openslug package symlinks? Jul 23 12:09:18 I guess it could be done in the master makefile, but the list isn't exactly short Jul 23 12:11:18 I just installed openslug and wanted to get an nfs-root running. So I did a turnup nfs -i nfsserver:/path/ and it copied all those files successfully to the server. The turnup-script doesn't complain about anything but in the end it doesn't modify the /etc/fstab only the fstab on the nfs-server, so there is no nfs-root after a reboot. Any idea? Jul 23 12:11:19 Can't we put it in a metapackage under /openembedded/packages/openslug-symlinks and have a link from openslug/oe-symlink to that dir? Jul 23 12:11:50 And unpack it from the makefile Jul 23 12:12:04 (so strike the last link suggestion) Jul 23 12:13:30 malfi: afaik it doesn't depend on the fstab on the flash at all Jul 23 12:14:05 NAiL: well, what could be the reasen then, that the nfs-share isn't mounted after a reboot? Jul 23 12:14:17 malfi, check if there is a new linuxrc file in / Jul 23 12:14:46 hanjo-log: okay, thanks Jul 23 12:15:49 under the assumption that / is still on flash Jul 23 12:16:29 I'm reflashing it with openslug, just to be sure, to have a clean installation. Jul 23 12:16:38 you should have a line starting with "exec '/boot/nfs' ...." Jul 23 12:17:20 the linuxrc would fail-back to a flash boot if there is any problem with the nfs, maybe that is what is happening in your case Jul 23 12:17:26 fall Jul 23 12:18:20 hanjo_: Probably, but it's kinda strange, because the turnup seemed to run successfully and I checked all the files on my nfs-server Jul 23 12:19:11 how are your permissions set-up in /etc/exports? Check that your IP did not change after the reboot Jul 23 12:20:16 IP is via dhcp is always the same Jul 23 12:20:47 exports is /serv/slung *(rw,insecure,no_root_squash,sync,nohide) Jul 23 12:21:41 OK, you have * there Jul 23 12:22:01 should be unsecure as possible ;) Jul 23 12:24:11 if you run out of ideas, you can try to launch a sniffer like ethereal and check what is going on Jul 23 12:24:17 good idea Jul 23 12:24:27 this /linuxrc is run by an initrd? Jul 23 12:25:24 i beleive so Jul 23 12:26:22 senneth: was it you who sent me a mail earlier? My mailserver has been down, so I haven't recieved anything. Jul 23 12:26:36 no i didn't Jul 23 12:26:51 ok :) Jul 23 12:29:28 well, I reflashed my nslu2, then run turnup init, reboot, turnup nfs -i soandso, reboot, the led turns red (should be early kernel stuff) then blinking amber but then it turns to solid amber and nothing happens anymore :-( Jul 23 12:30:15 malfi, what image are you using? Jul 23 12:30:17 ingeba-toandfrom: discussed the symlinks with jbowler, and he does have an alternate solution. Using BBFILES := in openslug.conf Jul 23 12:31:01 hanjo_: 9d256b1e16e3de9ef5d05e861f869ea0 openslug-2.0-beta.bin Jul 23 12:31:08 NAiL: I don't understand how that works, but if it means we get rid of svn for it I'm happy. Jul 23 12:31:24 ingeba-toandfrom: that's hopefully the result, yes :) Jul 23 12:31:53 that is quite old Jul 23 12:31:54 NAiL: You can add streamripper to the "updated and added symlink" list Jul 23 12:32:07 neat Jul 23 12:32:17 Need all my apps in place by Monday. Jul 23 12:32:41 hanjo_: where can I find a newer binary-image? Jul 23 12:33:13 I beleive you can't since berlios is still down Jul 23 12:33:48 hanjo_: I though that berlios only hosts the source, doesn't it? Jul 23 12:34:02 I can put the latest beta on a web for you to download Jul 23 12:34:09 that would be great Jul 23 12:37:32 Where do I report things on 2.3beta? Jul 23 12:37:41 berlios should only be down for bitbake - I have been able to retreive openslug stuff from it Jul 23 12:37:56 berlios still down? Jul 23 12:38:04 i should prepare a snapshot then Jul 23 12:38:11 igneba-toandfrom: in slugbug under '2.3-beta'... could you add that? Jul 23 12:38:20 mickeyl: let me check Jul 23 12:38:23 jbowler: my words exactly ;) Jul 23 12:39:07 mickeyl: svn: Can't connect to host 'svn.berlios.de': Connection refused Jul 23 12:39:33 jbowler: bummer. seems to be localized to anon access though. Jul 23 12:39:35 i'll prepare a snapü Jul 23 12:39:58 Ok, that explains why I can pull using svn+ssh - I should have tried that for bitbake... Jul 23 12:40:18 anyone with an account on loglibrary.com that can request logging for #openslug also? Jul 23 12:40:39 NAiL: would it help to have 2.4-beta too? A lot of the '2.0-beta' bugs were not in 2.0, rather they arose after 2.0 in the development tree. Jul 23 12:41:10 hmm.. should that be 2.4-beta or just "unstable"? Jul 23 12:42:16 2.4-beta, probably Jul 23 12:44:40 Something which changes with each release - 'unstable' doesn't tell us whether the bug was 1.12+, 2.0+ or 2.3+ Jul 23 12:44:52 exactly Jul 23 12:45:00 so 2.4-beta is fine Jul 23 12:45:25 Yes (if you do a build now it will be called 2.4-beta) Jul 23 12:45:37 ingeba-toandfrom: could you add 2.4-beta to slugbug as well? Jul 23 12:46:09 NAiL: Will do asap Jul 23 12:46:38 malfi: if you get a solid amber ready/status LED either the kernel has crashed (in which case sometimes it will probably be off) or there is something very wrong in /etc/init.d/rc Jul 23 12:47:53 2.3-beta and 2.4-beta added to slugbug Jul 23 12:53:25 jbowler, I've sent malfi a tested 2.3-beta binary Jul 23 12:53:31 the same problem with 2.3, clean flash, turnup init, reboot, turnup nfs -i ..., reboot, -> solid amber Jul 23 12:53:53 Can you log in (i.e. when the LED is amber) Jul 23 12:53:54 * malfi is firing up the packet-sniffer. Jul 23 12:54:12 For that matter, does the NSLU2 respond to pings? Jul 23 12:54:17 jbowler: I'm not even able to ping the nslu2. Jul 23 12:54:56 So (to summarise) the LED goes [reboot] - amber, red, flashing amber, solid amber - correct? Jul 23 12:55:04 exactly Jul 23 12:55:41 Then it's crashing, kernel oops, some time during the startup. Jul 23 12:56:18 Is there a /var/log/messages on the NFS root partition? For that matter, are there any changed files there? Jul 23 12:56:20 it must have something to do with the nfs, because with turnup memstick it worked flawless. Jul 23 12:56:50 # ls -l var/log Jul 23 12:56:50 total 0 Jul 23 12:56:50 -rw-rw-r-- 1 root root 0 Jul 23 21:47 lastlog Jul 23 12:56:51 -rw-rw-r-- 1 root root 0 Jul 23 21:47 wtmp Jul 23 12:56:53 that's all Jul 23 12:57:13 what is the sniffer saying? Jul 23 12:58:06 * malfi is now confused. Now it booted to the solid green led. But it's not using the nfs-root. Jul 23 12:58:42 There should be a file in the NFS root called .recovery Jul 23 12:58:51 That stops it trying to use that root. Jul 23 12:58:58 damn, I have to arpflood my switch in order to put it in hub-mode Jul 23 12:59:25 there is no such file called .recovery Jul 23 12:59:43 can't you run the sniffer on the server? Jul 23 13:00:56 Hum... then the first time it crashed and now it fails to even mount the nfs root - so it falls back to flash. Jul 23 13:01:33 okay, tcpdump is now running on the nfs-server, stay tuned ;) Jul 23 13:01:51 No, the first tiem it didn't mount the NFS root either, because /boot/disk does this test: Jul 23 13:01:53 if checkmount /mnt && :>/mnt/.recovery Jul 23 13:01:53 Jul 23 13:02:41 I.e. a partition can only be used if it looks ok and is writeable. Jul 23 13:07:53 first i have to figure out a tcpdump expressions to exclude the ssh-traffic to the server Jul 23 13:08:45 just be carefull, nfs traffic is by default UDP Jul 23 13:09:21 malfi: the simple way: tcpdump -l | grep -v ssh Jul 23 13:10:10 * malfi is now using tcpdump -ni eth1 not src port ssh and not dst port ssh Jul 23 13:11:03 hehe, I think I'll keep using my simple grep :-) Jul 23 13:11:38 22:10:40.746668 IP 192.168.1.19 > 192.168.1.1: icmp 336: 192.168.1.19 udp port 68 unreachable (19 is the nslu2 and 1 the nfs-server) Jul 23 13:12:10 that is harmless, isn't it? Jul 23 13:13:52 that is dhcp Jul 23 13:15:34 well, now it ended up with a solid amber again. Jul 23 13:15:49 strange Jul 23 13:16:36 are you using fixed ip in the turnup nfs -i line ? Jul 23 13:16:55 yes Jul 23 13:17:27 now it's up and running, the tcpdump-log doesn't show any try to mount the nfs-root Jul 23 13:19:27 Edit /boot/nfs and put the line 'exec 2>/boot.log >&2' at the top of the script (i.e. before the first command) - be careful, if you get it wrong you will have to reflash... Jul 23 13:20:01 http://extraball.sunsite.dk/notepad.php?ID=8607 Jul 23 13:21:35 dually:/serv/mp3/slung/' Jul 23 13:21:48 plese try with a fixed IP there Jul 23 13:22:06 and open your 68 TCP/UDP on the servere Jul 23 13:22:11 okay, i did that before, but I retry it with that bootlog Jul 23 13:23:36 I've never had any problems with names (rather than IP), but my DHCP settup is fully functional (i.e. it returns nameserver information). Jul 23 13:23:47 so does mine Jul 23 13:29:49 http://extraball.sunsite.dk/notepad.php?ID=8608 Jul 23 13:30:50 probably i should turn exec 2>/boot.log >&2 into append-mode? Jul 23 13:32:04 port 68 isn't filtered or something like that. How should I "open" that port? It seems as my dhcp-daemon isn't listening on that port. Jul 23 13:33:14 this is what I have in iptables Jul 23 13:33:20 -A RH-Lokkit-0-50-INPUT -p udp -m udp -s 0/0 --sport 67:68 -d 0/0 --dport 67:68 -i eth0 -j ACCEPT ~ Jul 23 13:33:47 this is behind 2 firewalls, you might want to be more restrictive Jul 23 13:35:29 also, SIOCSIFNETMASK: Invalid argument, suggests that something is wrong with the netmask Jul 23 13:35:49 * malfi ponders. Jul 23 13:37:46 Yes, append mode helps - sorry, I forgot about that. Jul 23 13:37:47 oh, you're right, there is a typo Jul 23 13:37:57 subnet 192.168.1.0 netmask 225.255.255.0 { ...} Jul 23 13:38:13 * malfi blushes. Jul 23 13:38:38 That's fine isn't it? Jul 23 13:39:02 Oh.... 225 :) Jul 23 13:39:30 You can edit /etc/default/sysconf directly to fix it. Jul 23 13:40:21 isn't he using dhcp? Jul 23 13:40:47 that looks like it's from the dhcpd.conf on the server side Jul 23 13:41:12 yes, I fixed that now Jul 23 13:41:50 and modified exec 2>/boot.log >&2 to exec 2>>/boot.log >>&2 but now it stays solid red, so I bricked it once again :-/ Jul 23 13:42:07 the >>&2 part is invalid, isn't it? Jul 23 13:43:13 yes (it's 2>>/boot.log >&2) Jul 23 13:43:47 and the only recovery I know of is a reflash. Jul 23 13:44:12 yes, I'm doing it right now :) Jul 23 13:54:32 yeees Jul 23 13:54:52 :) Jul 23 13:55:02 now it's working, it was the typo in the dhcpd.conf Jul 23 13:55:14 thank you very much for your patience Jul 23 13:55:30 * malfi cheers. Jul 23 13:55:34 So an ssh in to the flash worked, even with the typo? Jul 23 13:55:41 yes Jul 23 13:56:08 i could do almost everything, ipkg update, ipkg install et cetera Jul 23 13:56:29 only mounting this nfs-root wasn't working Jul 23 13:56:33 netmask is important for proper broadcast, and rpc relies in part on using broadcast Jul 23 13:56:55 ic Jul 23 13:57:30 Ah, ok... malfi: can you update the wiki? It's important because turnup seems to work but then the system won't boot... Jul 23 13:57:33 okay, which packages do I need for native development? glibc, gcc, binutils, make, awk Jul 23 13:58:00 autotools like autoconf and automake aswell Jul 23 13:58:28 jbowler: I add a note, that a correct subnet-mask is crucial for nfs, anything else? Jul 23 13:59:12 is there a package with the kernel source (i.e. for building further modules...) Jul 23 13:59:20 That's it - something which says that it won't boot with a bad netmask. Jul 23 14:00:25 try kernel-dev Jul 23 14:02:09 malfi, I think that would sound too general Jul 23 14:02:52 I would suggest to put that " if there are problems..." add the line from jbolwer Jul 23 14:03:14 and if the log says SIOCSIFNETMASK: Invalid argument, then one should check the netmask Jul 23 14:04:23 DaKa2: something worth noting (though I don't think there is a solution), an ipkg update, ipkg upgrade sequence fails because the upgrade of busybox removes 'wget' and the following downloads fail. Jul 23 14:05:54 (This was on a system where I had already installed coreutils and grep). Repeating the 'ipkg upgrade' fixes it (apparently). Jul 23 14:06:13 I added "If you got any problems with a nfs root (turnup nfs ...) you can add a 'exec 2>>/boot.log >&2' to /boot/nfs and check /boot.log the next time it comes up. If it shows something like SIOCSIFNETMASK: Invalid argument, you probably have wrong subnetmask which results in broadcast failures and thus a nonworking nfs-root. With such a misconfiguration the nslu2 might hang while booting and showing a solid amber led." Jul 23 14:06:39 sounds good to me Jul 23 14:07:42 DaKa2: there's a bug in the coreutils [ link - it gets a \: Jul 23 14:07:43 lrwxrwxrwx 1 root root 17 Jul 23 2005 [ -> ../../bin/busybox Jul 23 14:07:44 -rwxr-xr-x 1 root root 23352 Jul 23 20:21 [.coreutils Jul 23 14:07:44 lrwxrwxrwx 1 root root 12 Jul 23 2005 \[ -> \[.coreutils Jul 23 14:09:00 jbowler: dammit.. Jul 23 14:10:49 I made a special case for that, and I thought it worked.. Jul 23 14:11:16 Bug 212, it's coreutils-5.1.3-r2 - is that the latest? Jul 23 14:11:35 yep.. Jul 23 14:12:59 Too much quoting. The "" alone is enough. Jul 23 14:13:08 no, it isn't.. Jul 23 14:13:19 update-alternatives --remove "[" "[.coreutils" Jul 23 14:13:20 sed: unterminated match expression Jul 23 14:13:35 it should be: update-alternatives --remove "[" "\[.coreutils", same for install.. Jul 23 14:13:38 I think Jul 23 14:13:47 Maybe.. Jul 23 14:14:39 oh.. just "[", or corse.. Jul 23 14:15:38 hm.. dont like this.. Jul 23 14:17:36 seems like you just cannot have a [ in the target part of update-alternatives.. Jul 23 14:17:45 busybox works, because it links to busybox.. Jul 23 14:18:52 I guess I have to rename [.coreutils to something else, or figure out why it wont work Jul 23 14:19:03 jbowler: any ideas? Jul 23 14:25:47 rename to bracketleft.coreutils? Jul 23 14:40:17 does anybody use several nslu2s with distcc or something like that? Jul 23 14:41:31 ah it's in the wiki, nevermind Jul 23 14:47:50 jbowler: ping ? Jul 23 15:18:42 oh well, a simple fix to coreutils would be: Jul 23 15:18:45 mv ${D}${bindir}/[ ${D}${bindir}/bracketleft.${PN} Jul 23 15:18:49 update-alternatives --install ${bindir}/"[" "[" bracketleft.${PN} 100 Jul 23 15:18:53 update-alternatives --remove "[" bracketleft.${PN} Jul 23 15:19:15 replacing the current [ lines Jul 23 15:20:39 jbowler: Im going to sleep, if you think that looks good, tell me and I can commit tomorrow, or if you feel like it maybe you could :-) Jul 23 15:23:54 I just sent a request for a loglibrary channel id for #openslug. So when gernika pops in here asking; Its okay. Jul 23 16:09:42 DaKa2: sorry, forgot to change my name Jul 23 16:10:21 DaKa2: the fix is to use test.coreutils as the target, not [.coreutils - since test and [ are the same and it is the base name of the executable which determines the behaviour (as in busybox). Jul 23 16:10:44 I'll try a fix. Jul 23 16:15:57 Something weird is going on. In the coreutils r1 build test and [ are different files (different inode and different size even), Jul 23 16:34:23 Ok, coreutils builds two versions of 'test.c' one from 'lbracket.c' #including test.c, truely gross. I've fixed the reported problem by mving [ to lbracket.coreutils. Jul 23 17:53:35 DaKa2: lbracket because that's the name of the .c source file (otherwise as your change). Jul 23 18:34:49 there is no kernel-source ipkg, is it? (just browsed http://ipkg.nslu2-linux.org/feeds/openslug/cross/2.4-beta/) Jul 23 18:40:47 no, to build a kernel you need to build from monotone Jul 23 18:41:27 malfi: there are no -src ipk for nslu2-linux - use the Master Makefile to build (www.nslu2-linux.org/Makefile) Jul 23 18:43:19 Or you can build from a release source tarball. Jul 23 19:16:20 okay thanks Jul 23 20:45:34 foo Jul 23 20:46:34 Thanks Genrika! Jul 23 20:46:55 openslug channel users, realtime logs available now at http://www.loglibrary.com/show_page/latest/182 **** ENDING LOGGING AT Sat Jul 23 23:59:56 2005