**** BEGIN LOGGING AT Tue Sep 23 03:00:00 2014 Sep 23 03:16:07 build #695 of ppc44x is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/695 Sep 23 06:18:50 build #664 of sibyte is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/664 Sep 23 06:39:36 wigyori r42650 trunk/target/linux/mxs/config-3.14 * mxs: disable mmc debugging Sep 23 06:43:06 wigyori r42651 trunk/target/linux/ mxs/patches-3.13 mxs/config-3.13 * mxs: remove 3.13 Sep 23 07:15:32 build #676 of avr32 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/676 Sep 23 07:44:57 build #741 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/741 Sep 23 08:39:04 There are several minor updates in telephony repository (which contains mostly small fixes and security fixes). Should I merge those changes into release for-14.07 branch? Sep 23 10:19:16 nbd r42652 trunk/target/linux/generic/files/drivers/net/phy/ar8216.c * ar71xx: ar8216: move policies, pvid to setup_port Sep 23 10:20:36 nbd r42653 trunk/target/linux/generic/files/drivers/net/phy/ar8216.c * ar71xx: ar8216: tagged+untagged on ar8327 (#12181) Sep 23 10:41:53 nbd r42654 trunk/ scripts/metadata.pl include/package-defaults.mk include/package-dumpinfo.mk scripts/metadata.pm * build: allow packages with build variants to explicitly select a default variant Sep 23 10:42:03 nbd r42655 trunk/package/libs/ustream-ssl/Makefile Sep 23 10:42:03 ustream-ssl: select polarssl as default variant, skip openssl/cyassl dependencies if unused Sep 23 11:02:47 build #727 of ramips is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/ramips/builds/727 Sep 23 11:14:17 build #194 of malta is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/malta/builds/194 Sep 23 11:22:47 <_trine> Download failed. Sep 23 11:22:47 <_trine> --2014-09-23 12:13:21-- http://mirror2.openwrt.org/sources/apr-util-1.5.3.tar.bz2 Sep 23 11:23:09 <_trine> its moved on to http://mirror2.openwrt.org/sources/apr-util-1.5.4.tar.bz2 Sep 23 11:23:20 <_trine> so this needs to be changed Sep 23 11:29:22 ok, I've fixed my edgerouter and was able to flash openwrt Sep 23 11:29:29 Devastator: good Sep 23 11:29:52 oddly it only shows br-lan, eth0, eth1, isn't it suppose to show eth2 as well? Sep 23 11:30:20 Devastator: maybe it uses a built-in switch for 2 of the 3 ports Sep 23 11:30:31 (just guessing) Sep 23 11:31:11 https://dev.openwrt.org/browser/trunk/target/linux/octeon/base-files/etc/uci-defaults/01_network#L16 Sep 23 11:31:59 hum.. so eth2 should be configured manually I guess?! Sep 23 11:32:09 ahh, that's another option :) Sep 23 11:32:19 ifconfig -a shows eth2 ? Sep 23 11:33:31 yup Sep 23 11:33:33 hahah Sep 23 11:33:35 nice! Sep 23 11:34:00 ok then I guess it's simply because OpenWrt just has lan and wan by default Sep 23 11:34:06 like on most consumer routers :) Sep 23 11:34:20 exactly Sep 23 11:34:34 mind doing some throughput tests on it ? Sep 23 11:34:45 routed :) Sep 23 11:40:56 sure, can you give me a test example? Sep 23 11:46:32 Devastator: pm ? Sep 23 11:46:51 stintel sure :) Sep 23 12:23:52 I guess I can remove a few of these packages, like kmod-ath, kmod-ath9k etc?! http://codepad.org/Fls0QYbu Sep 23 12:42:44 build #600 of iop32x is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/600 Sep 23 12:48:09 build #767 of at91 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/767 Sep 23 13:02:33 Devastator: yep, and mac80211/cfg80211/iw/libiwinfo* maybe Sep 23 13:02:39 unless they are dependency of luci Sep 23 13:36:08 stintel thanks! also, do you have access to the wiki? Sep 23 13:43:31 Devastator: I do, but you can just register an account Sep 23 13:43:36 I did so yesterday Sep 23 13:51:24 yeah, because current edgerouter lite installation instructions are a bit misleading Sep 23 15:11:56 I'm chasing an apparent uclibc sockets problem. Could use clues where to look Sep 23 15:12:10 let me try and explain the problem fully Sep 23 15:12:49 One end uses 'nc' to write then read a value. The other end is a standard BSD sockets server - bind, accept, etc. It does a read then writes the result then closes the socket. Sep 23 15:13:38 if the nc end is run under uclibc on openwrt, then "sometimes" nc hangs, apparently waiting upon a close notification. Sep 23 15:13:53 I've tried various cominations, putting the ends on my desktop ubuntu system, etc. It doesn't matter where the server is. Sep 23 15:14:11 The hang is more likely if the server does a non-blocking connection, but it can happen either way Sep 23 15:14:13 nbd, I posted in ath5k Sep 23 15:14:18 nbd, https://lists.ath5k.org/pipermail/ath5k-devel/2014-September/005650.html Sep 23 15:14:34 and often when it does close, it takes a noticable amount of time - maybe some hundreds of milliseconds Sep 23 15:14:38 nbd, in case you have an interest in it Sep 23 15:15:11 Has anyone seen anything like this, or want to suggest where to look? Sep 23 15:15:54 what is your platform ? Sep 23 15:15:58 what is the configuration Sep 23 15:16:57 CONFIG_TARGET_ikf71xx=y Sep 23 15:17:05 it's a vendor specific thing. MIPS big endian Sep 23 15:17:22 you want the .config? Sep 23 15:17:35 I do not have this platform Sep 23 15:17:42 so no Sep 23 15:18:04 But how long/big are your transfers Sep 23 15:18:12 and over which transport ? Sep 23 15:18:13 10 bytes or so Sep 23 15:18:15 TCP Sep 23 15:18:23 ethernet ? Sep 23 15:18:24 wifi ? Sep 23 15:18:30 how often ? Sep 23 15:18:31 ethernet or local Sep 23 15:18:42 once a second in the test Sep 23 15:18:51 this is not like a volume problem Sep 23 15:19:00 if you lower the interval Sep 23 15:19:02 what happens ? Sep 23 15:19:07 in a running system, it's local comms Sep 23 15:19:12 it hangs no matter what Sep 23 15:19:37 it'll eventually hang, and the test is over at that point Sep 23 15:20:30 I can provide the source and test script if you want Sep 23 15:20:44 I don't have another uclibc system to hand to test this. All my other openwrt systems are eglibc Sep 23 15:20:54 this is same code I have running on 5 other eglibc systems. Sep 23 15:20:55 I understand you have been using non openwrt platforms to try this test as well Sep 23 15:20:57 correct ? Sep 23 15:21:03 and ubuntu yes Sep 23 15:21:17 the "other eglibc systems" are openwrt Sep 23 15:25:38 anyone have any further ideas on how to try and narrow this down? Sep 23 15:25:44 crazy ideas welcome Sep 23 15:28:52 when you run your code on eglibc systems it works ? Sep 23 15:28:57 as I said Sep 23 15:29:03 but that's an oversimplication Sep 23 15:29:16 there's two ends: when the 'nc' end is on uclibc it doesn't work Sep 23 15:29:24 it doesn't matter where the server end is Sep 23 15:30:03 how do you use nc ? Sep 23 15:30:08 pastebin i Sep 23 15:30:09 pastebin it Sep 23 15:30:15 hold tight Sep 23 15:30:41 http://pastebin.com/LdJr4YtL server Sep 23 15:31:08 nc send is: echo -n "testing" | nc 9055 Sep 23 15:33:16 where did you compile your code ? Sep 23 15:33:20 and how Sep 23 15:33:34 on my ubuntu machine with the openwrt cross compiler. Where else? Sep 23 15:33:49 Just making sure Sep 23 15:33:52 ....-gcc server.c -Wall -o server Sep 23 15:35:45 (after changing line 39) Sep 23 15:36:14 indeed Sep 23 15:36:41 you said you were sending then receiving with nc? what's the client script? Sep 23 15:36:54 (and you said it didn't matter where the server was right?) Sep 23 15:36:54 nc is the "client script" Sep 23 15:37:04 you can put a shell while loop around it if you want Sep 23 15:37:08 right, doesn't matter Sep 23 15:37:15 oh, right, didn't notice that nc was returning with the reply :) Sep 23 15:37:19 :o Sep 23 15:37:53 but yes, thanks for checking. these kinds of details do matter. Sep 23 15:38:42 which nc? busybox? or another one? Sep 23 15:38:47 busybox nc Sep 23 15:38:53 my nc isn't even connecting, it's just saying connection refused :| Sep 23 15:39:06 local works, and netstat shows 0.0.0.0 as the listen address properly :| Sep 23 15:39:09 run the server first with the right port Sep 23 15:39:21 (obviously) Sep 23 15:40:14 oh, doh, my bad, used the wrong ip :) Sep 23 15:40:23 happens to the best of us Sep 23 15:40:44 I might have to cobble up another uclibc system here. Sep 23 15:40:50 how long's this meant to take to fail? Sep 23 15:40:56 maybe try it 10 times Sep 23 15:41:07 it's been running in a while truee loop for ... a minute now? Sep 23 15:41:10 hm Sep 23 15:41:26 that's on an aold ar2317, running the AA branch. Sep 23 15:41:27 so what's your platform? And is your nc busybox? Sep 23 15:41:32 yeah, nc from busybox Sep 23 15:41:38 what version of uclibc? Sep 23 15:41:55 I could try BB, but it's not set up right now, was doing other things. Sep 23 15:42:36 actrually, never mind, more user error, running it on the right remote system, Sep 23 15:42:42 :o Sep 23 15:42:43 but it's still running now on the real system too Sep 23 15:42:48 right Sep 23 15:42:51 well, that's a data point Sep 23 15:43:11 wonder if there's something jacked in the vendor kernel Sep 23 15:43:13 hang on, trying to get the uclibc version Sep 23 15:43:29 CONFIG_UCLIBC_VERSION_0_9_33=y Sep 23 15:43:32 right Sep 23 15:44:08 it's meant to just hang on the client end right? Sep 23 15:44:13 yeah Sep 23 15:44:33 (it's easier to see if it's hung if you return an incrementing number, so that my screen doesn't just show Hello! statically) Sep 23 15:44:37 yes Sep 23 15:44:42 "you've done this before" Sep 23 15:44:46 but yeah, this is still just running fine on the atheros board. Sep 23 15:44:50 yes, I have :) Sep 23 15:45:01 no idea about your problem though. Sep 23 15:45:17 CONFIG_UCLIBC_VERSION="0.9.33.2" Sep 23 15:45:23 is what is also in the .config Sep 23 15:46:23 well, thanks. Let me try some stuff Sep 23 15:46:32 I want to verify FIN with tcpdump, and try the non-bb nc Sep 23 15:47:26 hm, not sure the latter is in openwrt Sep 23 15:51:05 "netcat" ? Sep 23 15:51:13 it's on my BB feeds at least Sep 23 15:51:22 also ncat too. Sep 23 16:05:25 * Chocks looks more Sep 23 16:24:34 still happens with gnu nc Sep 23 16:24:42 which isn't too much of a surprise, I saw this outside of that Sep 23 16:44:57 presumably the real situation wasn't netcat, but another socket process of your own? Sep 23 16:45:20 yes sir Sep 23 16:45:31 it's actually in PHP, but I wanted to rule out problems there Sep 23 16:45:49 actually, not quite true, I have a load of this type of stuff going on, and some of it is in fact nc Sep 23 16:46:03 the point of course was to make as simple test as possible Sep 23 16:46:16 vendor claims they can't reproduce the problem. Sep 23 16:46:33 I can see there's some delay in delivery of TCP packets when it gets stuck Sep 23 16:51:29 mine's still running... Sep 23 16:51:36 right Sep 23 16:51:48 I'm going to reload their latest image, and try again Sep 23 16:51:56 of course, being vendors, it's "not their problem" Sep 23 17:16:15 heh, AA has newer libevent2 than my fedora desktop Sep 23 18:02:08 build #187 of adm8668 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/adm8668/builds/187 Sep 23 19:45:58 build #586 of mcs814x is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/mcs814x/builds/586 Sep 23 20:05:05 cyrusff: around? Sep 23 20:07:16 ichilton: sort of ;) Sep 23 20:07:31 cyrusff: are you the expert on dhcpv6? :) Sep 23 20:08:24 well i've implemented a server and a client, don't know if that makes me an expert Sep 23 20:08:30 but ask away Sep 23 20:08:30 haha Sep 23 20:08:51 I'm using 14.07-rc3 - so the new native v6 stuff Sep 23 20:09:06 is there an advantage to using either ra, dhcpv6, or both? Sep 23 20:09:47 well RAs have the advantage of being effective instantaenously Sep 23 20:09:56 i.e. the router can send you updates every time Sep 23 20:10:07 every time what? Sep 23 20:10:15 everytime something changes Sep 23 20:11:24 let's say your ISP connection was dropped and reestablished Sep 23 20:11:30 and for some reason the public addresses changed Sep 23 20:11:43 router can just send an RA and the clients will act on it Sep 23 20:12:00 ah right - ra would give your network new ips instantly - dhcp would only be next time they get a lease? Sep 23 20:12:35 right with dhcp clients only poll based on the lease-time. there is a push-mechanism however most clients don't seem to implement it Sep 23 20:12:39 however there is a catch Sep 23 20:13:05 windows e.g. doesn't evaluate DNS-servers sent along RAs Sep 23 20:13:28 so it always needs to use at least stateless DHCPv6 (without addresses) or stateful along with RAs Sep 23 20:13:43 to get to know the dns-servers Sep 23 20:13:49 ok Sep 23 20:14:09 so what's the score with setting both dhcpv6 and ra to 'server'? Sep 23 20:14:14 how do the 2 work together? Sep 23 20:14:53 well the other pro for dhcpv6 is that you know which clients are "connected" to your router in a way and some of them report a hostname via dhcpv6 as well Sep 23 20:14:55 well Sep 23 20:15:07 clients can use both RAs and DHCPv6 simultaneously Sep 23 20:15:13 and combine information from both Sep 23 20:15:28 so you can sort of combine the best of both worlds Sep 23 20:15:29 this conversation should go in the wiki. great info here :) Sep 23 20:15:38 does that mean they'll end up with multiple ip's - one from each? Sep 23 20:15:42 yes Sep 23 20:15:47 ok Sep 23 20:15:56 and it's the client's job to choose which to use? Sep 23 20:16:03 (as the default outgoing) Sep 23 20:16:16 however clients cannot use DHCPv6 exclusively since you cannot configure routes using that (you need RAs here) Sep 23 20:16:47 yeah IPv6 clients have a fancy source-address selection algorithm Sep 23 20:16:47 ah, so you can't actually do dhcpv6 server and ra disabled? Sep 23 20:16:53 right Sep 23 20:16:58 ok Sep 23 20:17:18 because then you neither have a default-route nor could clients talk to other clients on the same link Sep 23 20:17:50 so both on is recommended? Sep 23 20:17:54 yes Sep 23 20:17:56 k Sep 23 20:18:10 you can tune the value management_level which will affect how clients get addresses Sep 23 20:18:12 AFAIK clients will use the most specific prefix as the default outgoing address Sep 23 20:18:24 and how do these mechanisms generate the ip to give it? Sep 23 20:19:21 - and related, waht if I want to give a specific client a specific, short address - like 2001:xxx:xxx:1::10, rather than a long auto generated one? Sep 23 20:19:28 heffer: there is a 30 page rfc on the selection (RFC 6724) itsnot that easy to summarize it in a sentence Sep 23 20:19:41 ichilton: that you can do with dhcpv6 Sep 23 20:20:03 RAs will usually result in random host-ids (lower 64-bits of the address) Sep 23 20:20:38 most dhcpv6 servers support static leases similat to those you know from dhcpv4 Sep 23 20:20:51 but both derive it from the mac address? Sep 23 20:21:23 for RAs they are mostly random now for clients (not based on mac-address) Sep 23 20:21:33 and clients even change them from time to time Sep 23 20:21:41 thats called "privacy extensions" Sep 23 20:21:48 ah yes Sep 23 20:21:50 for dhcpv6 its totally up to the server what addresses Sep 23 20:21:55 clients will receive Sep 23 20:22:24 right now, I only have dhcpv6 on, and I seem to get 2x ip's to each client - one being temporary and the one it seems to use by default Sep 23 20:22:28 and one based on the mac Sep 23 20:22:37 is that the client requesting 2? Sep 23 20:23:05 I am gonna love this irc channel, many valuable info :) openwrt rocks! Sep 23 20:23:05 well acting on the RA alone clients sometimes derive 2 Sep 23 20:23:14 one based on the mac-address Sep 23 20:23:24 (which they dont usually use for outgoing traffic) Sep 23 20:23:37 so if I turn on ra and dhcpv6, will I get 3 addresses? - an ra one, a dhcpv6 one and a temporary one? Sep 23 20:23:40 then one randomly assigned and sometimes changing (privacy / temporary address) Sep 23 20:23:57 ichilton: yeah depends on the client implementation you will get either 2 or 3 Sep 23 20:24:16 incidently, is it possible to get something like openssh to use the static one rather than the temporary one for outgoing? Sep 23 20:24:26 multiplied by the number of prefixes you have (usually one per ISP + maybe one local prefix) Sep 23 20:24:56 sure applications can choose whatever address of those they want Sep 23 20:25:12 if they expose that as a command-line flag is the other question Sep 23 20:25:28 ssh -b desired-source-ipv6 ... Sep 23 20:25:36 there you go ;) Sep 23 20:26:33 ah, cool Sep 23 20:26:57 cyrusff: how are the dhcpv6 ipv6 addresses generated in case there is no static lease for a requesting client? Sep 23 20:27:01 ah, that was the other odd thing - the global ula_prefix Sep 23 20:27:19 cyrusff: completely random, sequentially? Sep 23 20:27:55 ^ that was one of my questions also :) Sep 23 20:28:24 jow_laptop: in openwrt the duid (which is based on the mac) is used to seed an rng. and then its simply random until it hits a free one Sep 23 20:28:38 i think thats sort of similar to what dnsmasq does Sep 23 20:28:53 but in openwrt i only use the lower 3 bytes as a default pool Sep 23 20:29:04 so addresses don't get overly complicated Sep 23 20:29:23 thats something we should document Sep 23 20:29:38 or no even the lower 3 nibbles i think Sep 23 20:30:01 it would be cool if the bounds could be configured Sep 23 20:30:20 yeah guess i should add an option Sep 23 20:30:33 but static leases can already be outside of these bounds Sep 23 20:31:09 thats fine because they're user administered anyway Sep 23 20:32:37 so you're saying it's pure coincidence that my laptop's mac ends in ba:fd:72 and it's current non-temporary address ends in: feba:fd72 ? Sep 23 20:33:27 ichilton: no thats one addressed the host derived from the RA Sep 23 20:33:37 host being client? Sep 23 20:33:40 yep Sep 23 20:33:44 ah, ok Sep 23 20:33:55 that's odd, as i've got ra turned off Sep 23 20:33:56 oh Sep 23 20:34:10 i've got ra commented - does it default to server and you have to set it to disabled to turn it off? Sep 23 20:34:19 not really Sep 23 20:34:37 hmm, so in theory i'm currently dhcpv6 only... Sep 23 20:34:47 maybe you didn't apply the config change? Sep 23 20:34:57 e.g. /etc/init.d/odhcpd reload Sep 23 20:35:06 oh - I thought it was part of dnsmasq? Sep 23 20:35:10 no Sep 23 20:35:13 I was just restarting dnsmasq Sep 23 20:35:37 oops Sep 23 20:35:40 hehe Sep 23 20:35:57 well dnsmasq does have dhcpv6 support but its lacking bit behind on some ends Sep 23 20:35:59 so odhcpd is doing all dhcp - v4 & v6 - and RA? Sep 23 20:36:07 no dnsmasq is doing v4 Sep 23 20:36:11 ah, ok Sep 23 20:36:32 you could in theory use either one of those for all Sep 23 20:36:36 so dnsmasq is doing dns caching and dhcp v4 and odhcpd is doing dhcpv6 and RA? Sep 23 20:36:40 right Sep 23 20:36:45 they just both use /etc/config/dhcp? Sep 23 20:36:48 yep Sep 23 20:36:50 k Sep 23 20:36:58 maybe we'll resort to either one at some point Sep 23 20:37:03 the global ula_prefix..... Sep 23 20:37:20 i've got a /48 from my isp - should I just set that to blank? Sep 23 20:37:46 or use list ip6class wan6 on each network in /etc/config/dhcp? Sep 23 20:37:52 s/network/interface/ Sep 23 20:38:08 ip6class doesn't do anything in /etc/config/dhcp Sep 23 20:38:17 its only used in config/network Sep 23 20:38:26 sorry, in config/network Sep 23 20:38:53 yeah you could do that or just remove the ula_prefix if you don't need it Sep 23 20:39:03 what's the point of it? Sep 23 20:39:18 or story behind it.. Sep 23 20:39:23 its mainly there to provide ipv6-connectivity if there is no ipv6-uplink or if the uplink is temporarily broken for some reason Sep 23 20:39:56 or if you want stable addresses for internal use because the isp changes your public ipv6-prefix from time to time Sep 23 20:40:27 think of printers, servers, multimedia appliances Sep 23 20:41:27 ah, gotcha Sep 23 20:41:44 so it's potentially worth leaving in, so you have some private addresses Sep 23 20:42:09 well at least it shouldn't break anything Sep 23 20:42:10 and the clients will just use internet routable ones by default anyway Sep 23 20:42:31 does it just store the leases on those in memory though? Sep 23 20:42:40 i.e if your router reboots, all hosts will get different ones? Sep 23 20:42:43 or is it persistant? Sep 23 20:42:55 ra based addrs always end up being the same Sep 23 20:43:06 because it's based on the mac? Sep 23 20:43:07 jow_laptop: well except the privacy ones Sep 23 20:43:09 ;) Sep 23 20:43:12 yes Sep 23 20:43:23 dhcpv6 ones are stored if you added them statically Sep 23 20:43:23 but what about dhcpv6 ones? Sep 23 20:43:34 as in if you put in /etc/config/dhcp? Sep 23 20:43:35 the dhcpv6 based ones are generated using a seeded prng Sep 23 20:43:58 so they're random but the random result is always the same for a same requesting mac Sep 23 20:44:01 in parctise Sep 23 20:44:12 ah - even across reboots? Sep 23 20:44:50 thats what I understood Sep 23 20:44:54 yep Sep 23 20:45:01 so apart from the temporary/privacy addresses, whether it gets it from dhcpv6 or ra - they'll always be the same for a given mac, even across router reboots - correct? Sep 23 20:45:10 yes Sep 23 20:45:14 brill, thanks Sep 23 20:45:19 unless you run into a rare collision case Sep 23 20:45:25 is there a way of telling if an ip on a client is from RA or dhcpv6? Sep 23 20:45:43 dhcpv6 ones have a prefix of /128 Sep 23 20:45:47 RAs usually have /64 Sep 23 20:46:30 oh Sep 23 20:46:47 how does a /128 work? - how does it see the gateway? Sep 23 20:47:09 well it just means that there is no on-link route Sep 23 20:47:27 do either of them contain any common characters? Sep 23 20:47:32 also RA based ones usually look like ...xxxx:xxff:fexx:xxxx Sep 23 20:47:41 ah, right - that's what I thought Sep 23 20:47:43 where the x are from your mac address Sep 23 20:47:59 if you only have a /128 then your client sends everything to the router Sep 23 20:48:08 even if the target would potentially be on the same link Sep 23 20:48:49 if there is at least one /64 for the same prefix then it sends traffic with destinaton on the same link to the target itself and not to the router Sep 23 20:51:28 some more explaination on the EUI-64 format (the last 64 bits used for a RA derived address): http://howdoesinternetwork.com/2013/slaac-ipv6-stateless-address-autoconfiguration Sep 23 20:55:41 Thank you very much to you both for all the answers/info :) Sep 23 20:55:45 Most helpful! Sep 23 20:58:57 jow_laptop: long shot, but things like ucix, any chance of that sort of thing ever moving into libuci proper? (http://nbd.name/gitweb.cgi?p=luci.git;a=blob;f=contrib/fwd/src/ucix.c) Sep 23 20:59:51 karlp: I don't think so Sep 23 21:01:45 is the future for client apps to use ubus rpc instead of libuci? or just want to keep libuci as small as possible? Sep 23 21:03:11 I'm not sure if local apps will need to go via the rpc i nthe future? Sep 23 21:03:40 erm s/?/./ Sep 23 21:05:05 no problem, was just loading some uci config files this evening. Sep 23 21:05:37 I already have a few ucix style things of my own, bt was lookign for the foreach_section() type things that I'd not used before, found the ucix stuff, Sep 23 21:06:00 it's implemented in two different luci contrib things that I found at least. Sep 23 21:06:01 yeah a callback interator in the standard api would be nice Sep 23 21:06:19 doesn't need to be all of the ucix stuff or anything, but a few more helpers wouldn't hurt. Sep 23 21:06:33 (is there any doxygen/similar docs built anywher) Sep 23 21:07:28 hm, don't think so - I always grepped the info from the headers Sep 23 21:07:38 but a LXR would be a good idea I guess Sep 23 21:07:57 for libubox, ubus, libuci, ... Sep 23 21:08:30 is that anything I can help with? or is it just a matter of server admin work? Sep 23 21:08:36 (because, yes, it would be awesome) Sep 23 21:09:18 I just looked at lxr Sep 23 21:09:36 it seems to be a self contained perlscript with a bunch of assets for the generated html Sep 23 22:10:44 cyrusff: you said you could do a static lease with dhcpv6 - do you do that just the same as v4? Sep 23 22:11:00 how do you give it a v4 and v6 to assign? Sep 23 22:15:20 karlp: http://lxr.mein.io/ Sep 23 22:15:37 karlp: not sure if the dns is propagated yet Sep 23 22:15:53 set up lxr for a few openwrt projects Sep 23 22:42:02 build #737 of cobalt is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/737 Sep 23 22:49:51 a question about openwrt in general and another about openwrt on edgerouter lite: 1- is there a way to use iptables-restore for loading firewall rules in openwrt? I've found a 7 year old ticket that is marked as closed but I didn't understand if it was included or not Sep 23 22:51:12 the second question is: edgerouter lite is suppose to have close to 2GB of space, although when I type df -h, / is with 43MB, is there a way to take advantage of the rest of space? Sep 23 23:15:46 CONFIG_TARGET_ROOTFS_PARTSIZE=48 Sep 23 23:22:04 swalker so I have to modify that to 1500 or something? Sep 23 23:22:46 yes Sep 23 23:23:13 interesting Sep 23 23:23:58 I can imagine why this is set this way, because octeon only has a generic board profile Sep 23 23:40:47 <_trine> Devastator, is the firmware on that edge router not okay as it is Sep 23 23:41:43 well, it is, it's just a few details that I want to work with Sep 23 23:42:37 because there is only a generic profile, it installs ath9k on this router that doesn't have wireless Sep 23 23:42:58 I will try to create some RFCs and see how it goes Sep 23 23:43:41 <_trine> I have spent all day exploring how to jtag my broken dockstar Sep 23 23:43:59 any progress? Sep 23 23:44:07 <_trine> I finally succeeded earlier Sep 23 23:44:23 <_trine> it is now up and running again Sep 23 23:44:53 <_trine> it means dockstar are now unbrickable Sep 23 23:45:11 great! now you can test php :P Sep 23 23:45:38 <_trine> php is broken but there is a workaround Sep 23 23:46:37 * _trine wonders if there is more broken packages than working ones in openwrt at the moment Sep 23 23:46:41 <_trine> :P Sep 23 23:47:13 <_trine> reaver does not work either Sep 23 23:47:27 <_trine> its something odd about mon0 Sep 23 23:49:16 you have to understand, and you probably do, openwrt went through a major change on this release Sep 23 23:50:29 packages feed needed some love, and that's what they are trying to give now, if someone needs/uses a package, then find ways to maintain it or help maintaining Sep 23 23:50:48 in my opinion this is fair Sep 23 23:52:10 <_trine> well you are correct in what you say,, but it still doesn't alter the fact that there is much brokeness Sep 23 23:54:31 yes, but this is how things improve, it is also a way to detect what is broken Sep 23 23:55:06 in a few months things will be smoother and solid Sep 23 23:55:21 with at least a maintainer to complain at Sep 23 23:58:03 <_trine> hmmm Sep 24 02:33:54 build #192 of pxa is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/pxa/builds/192 **** ENDING LOGGING AT Wed Sep 24 03:00:00 2014