**** BEGIN LOGGING AT Sun Jun 30 03:00:18 2019 Jun 30 11:16:12 angelsl: the patches were posted 2 days ago on the wireless list i think Jun 30 11:18:17 oh yeah, i saw them Jun 30 11:19:12 ldir: could you help me test a fq_codel patch dave sent me ? Jun 30 18:28:35 hi all. been away a while (busy with work in a new job). anyway, just rebased and did "make defconfig oldconfig" and I'm seeing: Jun 30 18:28:48 tmp/.config-package.in:55: syntax error Jun 30 18:28:49 tmp/.config-package.in:54: invalid option Jun 30 18:29:25 etc. the lines in question look innocuous. what am I missing? Jun 30 18:30:01 54 select PACKAGE_ucert if PACKAGE_ucert-full 55 select PACKAGE_libblobmsg-json Jun 30 18:33:11 hmmm… all the failing lines have "select" or "default" with if x Hi people just got this tweet on the openwrt twitter i run. @OpenwrtH Any News on Pie SQM coming to OpenWRT? Jun 30 18:35:40 I googled pie sqm and cant find out what it is and how it difers from cake Jun 30 18:35:48 differs* Jun 30 18:36:33 philipp64 Hi mate sorry I don't know what's going on with your make file, but congrats on the new job. Jun 30 18:37:16 philipp64: make -C scripts/config clean Jun 30 18:37:22 and welcome back! Jun 30 18:40:25 ldir Do you know what pie SQM is? Jun 30 18:42:45 Tapper: https://tools.ietf.org/html/draft-ietf-aqm-pie-10 Jun 30 18:44:20 blogic: I'm a bit busy at the moment battling on multiple fronts - opening another one on Dave patches doesn't appeal at present time. Jun 30 18:45:02 :) Jun 30 18:45:36 ldir thanks for the link. Is pie newer than cake? Jun 30 18:47:03 Borromini you know anything about it mate? Jun 30 18:47:22 Tapper: no idea, not my area of expertise Jun 30 18:47:36 is pie even in OpenWrt? I have never hird of it. ldir OK mate. Jun 30 18:47:50 occam's razor would imply it is newer, since cake is integrated and pie isn't Jun 30 18:48:00 but that's just my take :) Jun 30 18:48:13 :) Jun 30 18:55:51 PIE is part of the DOCSIS v3 (?) cable modem spec - can't remember if in kernel. Not as clever as cake. Jun 30 18:58:52 pie in kernel since 2014 Jun 30 19:04:23 why would someone be asking for pie then i wonder Jun 30 19:06:32 Thanks ldir Jun 30 19:07:18 Borromini from what I read if I am getting it rite Pie is like cake but not as smart and uses less CPU Jun 30 19:09:25 I now have the OpenWrt twitter up to 92 followers Jun 30 19:47:56 neat Jun 30 20:06:35 Tapper: thanks. I'm OS Security Architect for Gigamon. interesting work. Jun 30 20:08:31 I just googled Gigamon intresting! Jun 30 20:08:33 especially now that the APT's are stepping it up… Jun 30 20:08:43 philipp64: sounds like a digimon ;p Jun 30 20:08:45 I had never herd of them before Jun 30 20:09:12 lol like a Pokemon Jun 30 20:10:28 sounds like a cool gig tho. Jun 30 20:10:33 It's a very specialized market… "network visibility" is the ability to confirm that what you see on your network is what you think (or believe) your network should be doing. oftentimes networks grow too quickly, new things get plugged in, etc. and we lose track of an accurate understanding of the network. network visibility devices help with seeing what's actually happening. Jun 30 20:10:45 * Tapper nods Jun 30 20:11:20 name came from Gigabit Monitoring mashup. Jun 30 20:11:49 philipp64 So I could not get on to a network and make it look like I am just a AP or some thing like that? Jun 30 20:11:51 But I think of Pokemon's too. Jun 30 20:12:02 Can't remember what Digimon makes… Jun 30 20:12:33 I think that Digimon was a off shout from Pokemon Jun 30 20:12:47 offshoot* Jun 30 20:13:30 ah. makes sense. Jun 30 20:14:56 it may have predated pokemon, not sure Jun 30 20:15:07 um… to your question, imagine making an inline switch and then mirroring "interesting" packets out an uplink for analysis by an IDS… Jun 30 20:16:14 as well as being able to do front-end decryption for HTTPS servers (or anything SSL-based), etc. and then do deep-packet inspection on the traffic as it goes by. Jun 30 20:17:12 anyone know uxgood's IRC handle? Jun 30 20:18:02 philipp64: do you use OpenWrt on these devices? Jun 30 20:18:36 No. CentOS. Jun 30 20:19:08 ok Jun 30 20:19:12 Though we're looking for anyone with Yocto experience… Jun 30 20:21:16 Contact me on LinkedIn for details… Jun 30 20:34:36 Tapper: the likes of codel, fq_codel, pie are queue management tools. Put simply they all shoot packets at some point if their internal queue/s get too long. If you're trying to control rate of flow of data into a downstream device that itself suffers from bufferbloat then you need a shaper to rate limit the release of packets from the queue/s Jun 30 20:35:51 cake combines queue management, with flow and host awareness/fairness, with a shaper that is also packet overhead aware. Jun 30 20:38:10 Hauke: has nft_reject_ipv4.ko been renamed to nf_reject_ipv4.ko in 4.19? does that affect include/netfilter.mk? Jun 30 20:38:54 the queue management bit is a hybrid of codel & blue. The shaper is also clever in that it uses 'deficit round robin'. Don't ask me to explain... I can't :-) Jun 30 20:40:16 I'm unaware of any other advanced queue management/shaper hybrid that is flow & per IP address (with NAT if done on the same box) capable. Jun 30 20:41:56 It does a lot of things which make it CPU expensive BUT if you put all the features together of say PIE aqm with a shaper, along with some things that nothing else *can* do then you'd be eating CPU with those too. Jun 30 20:42:39 and with that ldir goes to sleep Jun 30 20:48:24 philipp64: I think it didn't change much Jun 30 20:48:32 I saw this change regarding the dependecies. git.kernel.org/linus/a6615743704fdc179e227f84b7903edd1f0b4241 Jun 30 20:48:36 https://git.kernel.org/linus/a6615743704fdc179e227f84b7903edd1f0b4241 Jun 30 21:24:36 Hauke: seeing http://paste.ubuntu.com/p/F3gMdk2276/ after going from 4.14 to 4.19… Jun 30 21:29:57 seeing the following .ko's: Jun 30 21:30:34 linux-x86_64/linux-4.19.56/net/ipv4/netfilter/nf_reject_ipv4.ko Jun 30 21:30:34 linux-x86_64/linux-4.19.56/net/netfilter/nft_reject.ko Jun 30 21:30:34 linux-x86_64/linux-4.19.56/net/ipv6/netfilter/nf_reject_ipv6.ko Jun 30 21:31:04 so looks like a naming issue… it's nf_reject_{ipv4,ipv6}.ko but it's nft_reject.ko … Jun 30 21:32:48 where in package/kernel/linux/modules/netfilter.mk do the paths get expanded? Jun 30 21:36:04 Don't know where KCONFIG_NF_REJECT gets set. Jun 30 22:16:35 Hauke: there's also net/ipv4/netfilter/nft_reject_ipv4.c and net/ipv4/netfilter/nf_reject_ipv4.c … the first one looks newer. Jun 30 23:52:22 updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html **** ENDING LOGGING AT Mon Jul 01 02:59:57 2019