**** BEGIN LOGGING AT Thu May 24 02:59:58 2012 May 24 04:29:39 swalker * r31853 /packages/net/gpsd/ (6 files in 2 dirs): May 24 04:29:39 [packages] gpsd: update to 3.6 May 24 04:29:39 * add staging prefix patch for the pkg-config wrapper variables May 24 04:29:39 * re-enable libusb support May 24 04:29:39 * add a gpsd-clients package (#7879) May 24 04:29:39 * drop obsolete patches May 24 09:17:29 jow * r31854 /trunk/package/util-linux/Makefile: [package] util-linux: compile with -fPIC, some users of its libuuid.a require it, e.g. cryptsetup (#11508) May 24 10:22:04 I have some problem on 'http://luci.subsignal.org/trac/wiki/Documentation/JsonRpcHowTo' May 24 10:22:36 1. I can success get the TOKEN by : http://192.168.0.192/cgi-bin/luci/rpc/auth May 24 10:23:05 2. after get TOKEN i trigger another request: http://192.168.0.192/cgi-bin/luci/rpc/uci?auth=6b6ce42888db41692bd0ec217775e13e May 24 10:23:17 I always got: "HTTP request failed: The requested URL returned error: 403" May 24 10:23:36 I have install the luci-mod-rpc in OpenWrt. May 24 10:26:59 even the wrong password I still can get the TOKEN. :( May 24 10:34:29 build #24 of s3c24xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/24 May 24 13:13:52 build #17 of adm5120 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/17 May 24 13:39:36 is there anything that could limit reading uci config files too early in the boot process? May 24 13:39:51 I've got something that looks for an option in /e/c/system, May 24 13:40:17 and on a sysupgrade (keeping config files) it seems to fail to see the value in the config file May 24 15:28:09 I'm an idiot, I had a uci-defaults file that was running as well, and setting it to something else. May 24 15:41:06 nbd: ping & help please May 24 15:41:51 nbd: just prepared a firmware for tplink wr703n, it has the eth0 that i configured as WAN with pppoe and the wlan0 which should be br-lan.. but it doesnt bring up the br-lan .... May 24 15:42:22 you didn't omit swconfig right? May 24 15:42:37 yes i did, thanks, back to menuconfig :P May 24 15:42:41 and brctl May 24 15:42:48 I think that's bridge-utils May 24 15:43:07 i checked back kernel bridge support and busybox brctl but i forgot swconfig Olipro May 24 15:43:54 tho' it shouldnt be a switch right? its just one ether interface and one wireless interface.. May 24 15:53:25 dape: so br-lan only has wlan0? May 24 15:53:33 i think there might be a corner case there that i haven't fixed yet May 24 15:54:44 yes nbd. Olipro still br-lan wont come visible May 24 15:55:03 I want to periodically copy a file using cron from one router to another. But I want to avoid creating an ssh login for that (with public key), because shell access is not needed. Does anybody know a secure (encrypted) way to do that? May 24 15:55:11 dape: ok, i'll look into it May 24 15:55:33 you could make the shell not a shell but a script that does the task? May 24 15:55:58 nbd: this is my /etc/config/network -> http://codepad.org/YdVat6zP May 24 15:56:36 dape: i'm pretty sure i already know what the problem is May 24 15:56:47 oh, wait May 24 15:56:51 that config May 24 15:56:52 karlp: well I want to avoid the additional user login on the remote machine. Easiest would be to create a user and then scp the file with a public key mechanism. But that means that in theory it is possible to login with that key. I want to avoid that. May 24 15:56:52 maybe should i put option 'ifname' 'wlan0' in lan declaration ? May 24 15:56:55 isn't supposed to bring up br-lan May 24 15:56:56 at all May 24 15:57:06 it's an unbridged configuration May 24 15:57:16 yes, it has one ether and one wlan May 24 15:57:22 mb__: either the remote device fetches itself, or you're pushing May 24 15:57:28 if you're pushin, you need a login. May 24 15:57:32 i mean the bridge br-lan would be just wlan0 May 24 15:57:39 mb__: are you only copying a file, and not needed any action on it once copied ?? May 24 15:57:40 dape: you didn't configure a bridge May 24 15:58:00 nbd: well.. only wlan0 would be part of it.. May 24 15:58:05 what i mean is May 24 15:58:13 there's no "option type bridge" in the lan section May 24 15:58:21 meaning it's supposed to apply the configuration directly to wlan0 May 24 15:58:29 which is fine, given that you're not adding another interface May 24 15:58:38 OutBackDingo: Well I want the client machine just push a file and _not_ do any commands. The remote machine will then do the right thing (tm) with it. But I dunno if such a service already exists. May 24 15:59:10 but dnsmasq wont work since there is no br-lan, so i better just put option type 'bridge' under lan then May 24 15:59:52 mb__: probably writing a service would be best, say using like netcat May 24 16:00:24 mb__: rsync ? May 24 16:00:28 OutBackDingo: Hm, probably. But it needs to be encrypted. May 24 16:00:39 rsync works over ssh, but there we also have the login problem May 24 16:00:57 mb__: wget and a small http/php service on the other ap, but sounds like overkill May 24 16:01:34 Or does ssh not allow login, if /bin/false is specified as shell for the user? That would be an option then. But I dunno. May 24 16:01:45 id think netcats the most logical route May 24 16:01:49 nbd: hmm, even with option type 'bridge' and option 'ifname' 'wlan0' under config interface 'lan' i still dont have br-lan up and ofc dnsmasq wont work May 24 16:01:50 Would still allow pushing random files, though May 24 16:02:13 OutBackDingo: cool. Thanks for the advise :) May 24 16:02:48 http://www.terminally-incoherent.com/blog/2007/08/07/few-useful-netcat-tricks/ May 24 16:05:12 mb__: mb__here a better one maybe, using encryption May 24 16:05:13 OutBackDingo: Looks good. But it is unencrypted. May 24 16:05:14 http://bsdsupport.org/2007/03/q-how-do-i-encrypt-file-transfers-with-dd-and-netcat/ May 24 16:05:24 :) May 24 16:05:39 oh that looks interesting indeed. May 24 16:06:01 nbd: but if i do "wifi" then br-lan will come up and i can see the essid (so i guess both dnsmasq and hostapd are affected) May 24 16:06:20 OutBackDingo: So now it just additionally needs authentication :D May 24 16:07:11 mb__: basically, theres also cryptcat a crypto netcat port, though i dont believe its in packages, im looking at it now May 24 16:07:41 http://cryptcat.sourceforge.net/ May 24 16:16:10 dape: it's normal for the bridge to not be there before wlan0 gets brought up May 24 16:17:02 nbd: mkay, im not sure what to do next May 24 16:17:33 in /etc/config/wireless i told it to belong to network lan May 24 16:18:05 wireless has disabled 0 but it wont come up after boot, br-lan doesnt exist.. May 24 16:24:37 i just checked my old backup files when it was working, i think from december 2011, so my actual config looks like this -> http://codepad.org/AP2pgwS4 i edited USERNAME and PASSWORD May 24 16:24:57 nbd * r31855 /trunk/package/netifd/files/etc/init.d/network: netifd: bring up wifi at boot May 24 16:25:01 dape: try with that May 24 16:25:14 oki, thx May 24 16:34:03 nbd: doesnt work.. so i have bridge but not vlan support in kernel, i have swconfig, brctl but not vconfig and the configs above in dpaste May 24 16:35:22 nbd: "wifi" will start wlan0 and put 192.168.1.1 on it May 24 16:35:55 all of the above without option type bridge under config lan, let me try it with it May 24 16:36:24 nbd: with option type bridge in config lan WORKS, thanks! May 24 16:36:38 (tho' its a one interface bridge :P) May 24 16:59:56 jow_laptop: pcall should still work in luci right? I can see it used in some other places, but it doesn't seem to catch in mine? http://pastebin.com/7Xj8LnQB May 24 17:00:06 is that possibly an artifact of make runuhttpd? May 24 17:00:38 the path it spits out is not in [luci_dir]/host, it's finding the mqtt_library from my machine's usr/lib May 24 17:36:42 you're using it wrong May 24 17:37:03 pcall(func, arg1, arg2, ...) May 24 17:37:08 not May 24 17:37:15 pcall(func(arg1, arg2, ...)) May 24 17:40:39 the way you wrote it, pcall attempts to call the result of mqtt_client:connect() May 24 17:53:22 oh, duh. May 24 17:53:41 and then mqtt_client:connect() becomes pcall(mqtt_client.connect, May 24 17:53:46 not with the : May 24 18:25:11 nbd * r31856 /trunk/target/linux/generic/patches-3.3/621-sched_act_connmark.patch: kernel: fix a warning in the connmark module May 24 18:45:03 nbd * r31857 /trunk/package/netifd/Makefile: netifd: update to latest version, fixes overriding the mac address for bridge interfaces (#11526) May 24 19:29:55 oh bollocks, I knew I forgot something May 24 19:30:28 jow_laptop: so, I updated iptables (userspace) to the latest version and added support for -m rpfilter May 24 19:30:48 I had to remove a couple of nerfing patches to make it work, and adapt another May 24 19:31:00 but work it does May 24 20:37:34 Olipro: problem with current iptables is the merge of iptables and ip6tables, we didn't agree on a packagement strategy yet May 24 20:37:54 ah yes May 24 20:38:09 in my patch, it installs xtables-multi and symlinks to it May 24 20:39:00 busybox style May 24 20:39:07 yeah May 24 20:39:26 question is whether we give up on ip6tables and just ship multicall May 24 20:39:41 or whether we ship ip6tables as frontend for iptables which already contains the multicall May 24 20:40:07 or whether we build ip6tables statically and ship that while accepting the overhead in iptables May 24 20:40:21 why not just do it as above? install xtables-multi and make ip(6)tables a symlink to it May 24 20:40:38 if you call xtables-multi with no args, anything it lists you can create as a symlink May 24 20:40:57 because it unconditionally increases the size of installs which do not use ip6tables May 24 20:41:16 ah May 24 20:41:39 personally I wouldn't care because ip6tables is pretty much a necessity nowadays May 24 20:41:42 yeah, this is a situation where it'd arguably be preferable to actually give people not using IPv6 an older version of iptables May 24 20:42:00 but I can understand the people demanding the opposite May 24 20:42:59 at some point I think we should just remove the non-ipv6 option for binary releases May 24 20:43:26 people who really need a no-v6-support-compiled-in-variant can still build custom images with CONFIG_IPV6=n May 24 20:43:35 I'm definitely against avoiding upgrading a package for the benefit of those clinging onto older cruft May 24 20:43:58 in that respect, the buildenv could do with some version flexibility on packages May 24 20:44:28 if this var, download and compile this version with patch set X, otherwise, do this one with patch set Y May 24 20:44:30 we could offer an iptables and an iptables-noipv6 May 24 20:44:34 or something like that May 24 20:44:51 don't think there is a need to offer multiple versions May 24 20:44:54 yeah, that works May 24 20:45:20 well, you /could/ alter the latest version so that they're separate binaries May 24 20:45:28 but it'd be quite a pain in the arse May 24 20:46:02 that would kind of contradict the development direction upstream has taken May 24 20:46:04 shifting the current iptables in trunk to iptables-noipv6 and then making 1.4.13 plain ol' iptables would work May 24 20:46:10 they're working on merging since several releases May 24 20:46:19 while we would work on separating again May 24 20:46:56 if I find time I need to look at the size figures again May 24 20:47:41 "and then making 1.4.13 plain ol' iptables would work" <-- I meant the package name only May 24 20:47:51 so it'd be package/iptables May 24 20:47:58 you'd still get xtables-multi and all the symlinks May 24 21:01:29 i think if it's not too big, we should probably just put in the multi variant by default May 24 21:01:57 iptables changes more than some other packages, so our changes should not be too intrusive May 24 22:28:54 hello, not sure if this is the place to ask LuCI questions, but is there a way to make LuCI's httpclient asynchronous and not block when making requests? **** ENDING LOGGING AT Fri May 25 02:59:59 2012