**** BEGIN LOGGING AT Sat Nov 29 02:59:59 2014 Nov 29 07:47:10 build #806 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/806 Nov 29 10:32:29 luka r43425 trunk/target/linux/malta/config-3.12 * malta: drop 3.12 config file Nov 29 10:32:31 luka r43426 trunk/target/linux/malta/config-3.14 * malta: add support for 3.14 Nov 29 10:32:34 luka r43427 trunk/ (16 files in 2 dirs) * kernel: update 3.14 to 3.14.25 Nov 29 12:20:54 build #442 of mpc85xx is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/mpc85xx/builds/442 Nov 29 14:31:58 build #583 of octeon is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/octeon/builds/583 Nov 29 14:56:53 build #218 of omap is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/omap/builds/218 Nov 29 17:53:17 good morning folks Nov 29 17:53:38 I started playing a bit with the edgerouter lite last nite, built from scratch, fresh trunk Nov 29 17:53:44 and got some surprising results Nov 29 17:54:01 getting thruput in the 250mbit range, from lan to wan Nov 29 17:54:26 so now I'm wondering, does the current build use the hardware nat on that thing ? Nov 29 17:54:37 because these are very impressive numbers I'm getting, Nov 29 17:56:43 I did another test, raw thruput, fetching from the device itself, so no nat or routing involved Nov 29 17:57:07 and got a sustained 500mbit fetching 10 gigabyte files from a web server on it's upstream side using wget Nov 29 19:30:39 groz: no, offloading is not used Nov 29 19:30:48 groz: I've done some tests today as well, with edgeos Nov 29 19:31:18 curious what kind of numbers you were seeing Nov 29 19:31:30 my testing was a contrived network, set up to stress the box Nov 29 19:31:30 v1.6.0, lan->wan, offloading disabled: 170Mbps Nov 29 19:31:38 offloading enabled -> 940Mbps (!) Nov 29 19:31:49 2 clients both of which are gigabit units, with dual core processors Nov 29 19:31:57 then two servers on the other side, similar setup Nov 29 19:32:14 so both clients fetching 10 gigabyte files, one from each server Nov 29 19:32:39 and 940Mbps might have been the limit of thunderbolt ethernet adapter on the macbook Nov 29 19:32:58 yah, i was wondering if I was starting to bump into network limits Nov 29 19:33:11 did the same test, using a soekris 6501-50 as the router Nov 29 19:33:28 and it was able to sustain a tad over 350mbit Nov 29 19:33:45 now the edgeos 1.6 has kernel 3.10.20 Nov 29 19:34:03 and I suspect between that and 3.14 on openwrt, some more network improvements were added Nov 29 19:34:10 i'm not really interested in using the offload here, because there's a lot of things I cant do if we dump stuff into the offload Nov 29 19:34:16 I noticed an increase in speed on my rspro to, going from 3.10 -> 3.14 Nov 29 19:34:41 dont think we can enforce qos limits per client, if using offload Nov 29 19:34:52 I tried with fq_codel stuff added in v1.5.0 of edgeos and speeds dropped < 100Mbps Nov 29 19:34:53 but I could be mistaken Nov 29 19:35:49 so I just decided to disable qos/sqm on my router, and enabled some qos and shaping on my switch (modem is connected to the switch as well) Nov 29 19:36:45 i want to use qos on the router, so, am building a setup now where I can test using that Nov 29 19:36:48 since then an online speedtest gave 3-4Mbps higher downstream :P Nov 29 19:37:00 but, if the numbers I've seen today are any indication, this is a killer box for the price Nov 29 19:37:24 i've only got 100mbit internet here, so, doing tests online, limited by my isp Nov 29 19:37:36 I see, my internet is 200/10 Nov 29 19:37:45 is why I've contrived my network setup for testing Nov 29 19:38:00 yah, that's double mine, which is 100/5 Nov 29 19:38:01 and I expect the speed to go up even more in the coming few years Nov 29 19:38:23 our isp has been promising 250/10 for over a year, but, it's not here yet Nov 29 19:38:36 and since i'm a little ways out of town, probably be among the last to get it Nov 29 19:38:43 the CMTS I am on is going to be replaced somewhere next month Nov 29 19:38:50 but 100 is more than adaquate for what i do Nov 29 19:38:55 but that should just fix some IPv6 issues with the motorola bsr crap Nov 29 19:39:10 ipv6 is a non issue in my part of the world Nov 29 19:39:14 maybe little lower latency Nov 29 19:39:31 my ISP has enabled IPv6 for all customers with a home gateway Nov 29 19:39:38 is one of the first things I do here when configuring Nov 29 19:39:46 is remove everything to do with ipv6 Nov 29 19:39:49 heh Nov 29 19:39:56 so you have legacy Internet ;) Nov 29 19:39:58 where are you ? Nov 29 19:40:03 Belgium Nov 29 19:40:10 i'm on vancouver island, canadian west coast Nov 29 19:40:21 and 15km out of town Nov 29 19:40:32 so, 100/5 is pretty good considering where I am Nov 29 19:40:52 cable ? Nov 29 19:41:15 yes, i can get 100/5 on the cable, and the telco can deliver a whopping 6/512k here Nov 29 19:41:19 on a dsl line Nov 29 19:41:29 and if I was half a kilometer farther out Nov 29 19:41:33 then cable no longer available Nov 29 19:41:40 heh lucky Nov 29 19:41:44 planned Nov 29 19:41:45 6/0.5 is horrible Nov 29 19:41:58 when we bought this place, one of the things on our 'must have' list, was cable Nov 29 19:42:04 bought it last fall Nov 29 19:42:13 my phone has a faster connection Nov 29 19:42:23 I wonder why they still bother with DSL Nov 29 19:42:27 real estate agent tried to show us a few places farther out, but, I finally got it thru his thick skull Nov 29 19:42:40 check online first, and if the cable not available there, we aren't interested in even looking Nov 29 19:42:42 brb, door Nov 29 19:45:20 but I'm gonna do some more playing with this box, it's got a lot of potential Nov 29 19:45:57 got an electrian coming next week, gonna start putting fiber in the ground between the house and the observatory out on our back lot Nov 29 19:46:10 I've installed openwrt on the free space of the erl's usb stick Nov 29 19:46:11 so just toying with what my options are now Nov 29 19:46:17 and put another kernel on sda1 Nov 29 19:46:24 so I can choose what to boot in u-boot Nov 29 19:46:29 I cheated, opened it up, and put in a different usb stick Nov 29 19:46:42 so I still have the original sitting there with edgeos just in case Nov 29 19:46:51 but, I have zero interest in using edgeos Nov 29 19:46:51 just tested with openwrt, trunk build from yesterday, forwarding with nat: 322Mbps Nov 29 19:47:08 ya, that's in line with what I was seeing Nov 29 19:47:39 i had a burst up to 310, but sustained at just north of 250 Nov 29 19:47:52 I'll do a longer test Nov 29 19:48:04 this seems much faster actually than with edgeos and offloading disabled Nov 29 19:48:10 my sustained test was 20 gigabytes of data Nov 29 19:49:04 I need to put 3 computers out in the observatory, and a couple other ethernet connected gadgets Nov 29 19:49:19 contemplating the er pro cuz it's got the fiber ports already Nov 29 19:49:33 and, enough ethernet ports to handle everyting, just use it instead of a switch Nov 29 19:49:56 the heatsink gets pretty hot though, they should have put a slightly bigger one, and maybe small ones on the RAM chips too Nov 29 19:50:03 but I can add that myself :) Nov 29 19:50:26 i never checked that, maybe I should check it while under heavy load Nov 29 19:50:39 but, reality is, routers spend most of thier life sitting doing 'not much' Nov 29 19:50:48 then occaisionally push thru a burst of data Nov 29 19:51:06 well, consumer routers, yes :) Nov 29 19:51:17 mine is not warm right now, sitting on the desk, been doing 'not much' for a while Nov 29 19:51:22 consumer, small office, etc Nov 29 19:51:36 this is a 3 port box, not a backbone or data center type of thing Nov 29 19:54:35 build #227 of adm8668 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/adm8668/builds/227 Nov 29 19:55:16 heh Nov 29 19:56:08 doesn't look good: http://pastebin.com/S0DgmYFP Nov 29 19:57:24 build #759 of ppc44x is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/759 Nov 29 20:07:34 groz: are you running ext4 image on the ERL? and if yes, did you try sysupgrade? Nov 29 20:08:09 on phone Nov 29 20:08:27 back Nov 29 20:08:37 I actually fussed a bit with it Nov 29 20:09:06 i put block2mtd in kernel command line, then end up with more mtd, and have squash on the flash Nov 29 20:09:16 so it does rootfs + rootfs split stuff Nov 29 20:09:25 and ends up looking 'just like any other router' Nov 29 20:09:31 ok. I'm doing that on my openwrt vm Nov 29 20:09:35 and no, have not fussed with sysupgrade yet Nov 29 20:09:46 but I used the ext4 image on the ERL, and sysupgrade wiped the settings Nov 29 20:09:52 one of my goals, was to get it going without needing changes to uboot command lines Nov 29 20:10:04 yes, I looked at the sysupgrade scripts, very basic Nov 29 20:10:10 just explodes the tar, doesn't save config Nov 29 20:10:19 I haven't even looked at it farther yet Nov 29 20:10:25 i just started playing yesterday Nov 29 20:10:44 so i did a kernel_menuconfig and added a couple kernel command parameters Nov 29 20:10:45 my device arrived yesterday so same here Nov 29 20:12:10 block2mtd.block2mtd=/dev/sda2,65536,rootfs,20 root=/dev/mtdblock3 Nov 29 20:12:17 then dd the squash into /dev/sda2 Nov 29 20:12:31 and it does the normal firstboot and make jffs stuff Nov 29 20:12:46 i added that in the kernel command line configuration stuff Nov 29 20:12:55 rather than changing the on board uboot commands Nov 29 20:13:15 I wanted to leave uboot pristine, so it would still 'just work' if I put the original usb back in Nov 29 20:13:44 then i put a ramdisk kernel into the fat partition too Nov 29 20:13:55 ah I've not done saveenv, just setenv bootcmd = ... and then run bootcmd Nov 29 20:13:58 so, I can run it from uboot, end up with a ramdisk system Nov 29 20:14:07 and use that to re-write flash however I want Nov 29 20:14:31 i put in a new uboot env for a 'run from ramdisk' command Nov 29 20:14:48 the u-boot version is crazy old Nov 29 20:14:50 set a new variable, similar to bootcmd Nov 29 20:15:04 but it loads vmlinux.ramfs instead of vmlinux.64 Nov 29 20:15:11 I saved that one from uboot Nov 29 20:15:18 so, I can ctrl-c into ubbot Nov 29 20:15:24 then just type 'run ramdisk' Nov 29 20:15:29 and it boots my ramfs kernel Nov 29 20:15:53 it's a good way to play with various options, and, if I load a broken squashfs into the flash Nov 29 20:15:55 easy way out Nov 29 20:16:05 good idea Nov 29 20:17:02 my ramdisk command is the same as bootcmd, without the rootdelay and a different kernel name Nov 29 20:17:05 otherwise the same Nov 29 20:17:27 then I have that kernel in sda1 named vmlinux.ramfs Nov 29 20:17:46 once that one is tested, I wont change it, it's my back door to 'reflash it all easily' Nov 29 20:19:30 the fact that it's a simple usb stick is really handy too Nov 29 20:19:35 and to make things real easy Nov 29 20:19:51 I added a couple scripts in there that fetch a new kernel and rootfs from my build machine Nov 29 20:20:07 so, if I boot the ramdisk version, after mounting /dev/sda1 Nov 29 20:20:25 I can just run the scritps, they fetch fresh kernel and rootfs, stuff the rootfs into /dev/sda2 Nov 29 20:20:28 and restart it Nov 29 20:20:39 makes for fast turnaround when fussing with things Nov 29 20:20:53 and yes, I've built a few broken root file systems that dont start correctly Nov 29 20:21:06 was trivial easy to back out of that with this setup, so, it means I experiment more Nov 29 20:21:07 lol Nov 29 20:21:27 yeah I might steal that idea ;) Nov 29 20:22:20 cant steal it, cuz it was offered already for free Nov 29 20:22:25 so, no theft involved Nov 29 20:22:53 s/steal/copy/ Nov 29 20:22:59 :) Nov 29 20:23:28 take my word for it, you will be happy you did, the first time you build a rootfs that doesn't start Nov 29 20:23:34 cuz you dont have to pull the stick out Nov 29 20:23:43 saves a lot of hassle Nov 29 20:27:59 stintel, another trick, will possibly save you a lot of time Nov 29 20:28:24 if you want to use the block root stuff, it only works when there is rootfs and rootfs_data happening thru rootfs split Nov 29 20:28:35 but, if you bury the correct configs into the squashfs Nov 29 20:28:46 it does the rootfs split, then never touches the jffs Nov 29 20:28:54 and just goes on to use the block root partition Nov 29 20:29:10 thats how I got on the track of fussing with block2mtd Nov 29 20:29:16 and around that circle Nov 29 22:45:36 luka r43428 trunk/tools/mkimage/ (6 files in 2 dirs) * [tools] mkimage: update to 2014.10 Nov 29 23:47:30 luka r43429 trunk/package/ (10 files in 10 dirs) * [package] uboot-*: update license info in uboot packages Nov 30 00:21:09 kaloz r43430 packages/net/coova-chilli * coova-chilli: move to github Nov 30 01:15:13 build #722 of sibyte is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/722 Nov 30 01:15:48 luka r43431 trunk/target/ linux/generic/config-3.14 linux/generic/config-3.18 * generic: set CONFIG_GENERIC_NET_UTILS Nov 30 01:15:50 luka r43432 trunk/target/linux/kirkwood/config-3.14 * kirkwood: refresh kernel config Nov 30 01:15:53 luka r43433 trunk/package/ boot/uboot-envtools/Makefile boot/uboot-envtools/patches/001-compile.patch boot/uboot-envtools/patches/300-support-env-in-ubivol-chardev.patch * uboot-envtools: bump to 2014.10 Nov 30 01:22:14 luka r43434 trunk/target/linux/kirkwood/config-3.14 * kirkwood: change CRC32 implementation to CRC32_SLICEBY8 Nov 30 02:13:45 build #734 of avr32 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/734 **** ENDING LOGGING AT Sun Nov 30 02:59:58 2014