**** BEGIN LOGGING AT Sat Oct 10 02:59:59 2015 Oct 10 03:34:21 build #98 of ramips.mt7620 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.mt7620/builds/98 Oct 10 05:21:23 build #98 of ar71xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ar71xx/builds/98 Oct 10 08:11:36 build #114 of ramips is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/ramips/builds/114 Oct 10 12:05:55 <_trine> my router runs out of memory while downloading torrents but the memory never seems to get reclaimed later,, is this a memory leak in the kernel? Oct 10 12:27:21 _trine: nope Oct 10 12:36:22 luka r47167 trunk/package/boot/uboot-envtools/files/imx6 * uboot-envtools: Add Gateworks Ventana Support Oct 10 12:36:33 luka r47168 trunk/target/linux/ imx6/base-files/lib/upgrade/platform.sh imx6/base-files/etc/uci-defaults/02_network * imx6: fix platform checking Oct 10 12:44:26 <_trine> txomon, have you any idea why the memory never gets reclaimed Oct 10 12:44:58 _trine: can it be reclaimed? Where is the torrents? is it using /tmp? Oct 10 12:45:23 <_trine> on my SSD Oct 10 12:45:25 s/the torr/the device downloading the torr/ Oct 10 12:45:37 _trine: does it cache in /tmp? Oct 10 12:45:43 or any tmpfs/ramfs? Oct 10 12:45:52 <_trine> I dont think so Oct 10 12:46:18 if you kill the program, and doesn't show anymore in ps, where is that memory profiled? Oct 10 12:46:45 <_trine> dont know Oct 10 12:47:45 <_trine> this is in /tmp -rw------- 1 root root 344064 Oct 10 11:34 dynamic_dns_upd.1444476872.2326.10.core Oct 10 12:48:18 <_trine> and Oct 10 12:48:21 <_trine> -rw------- 1 root root 65469 Oct 7 11:29 luci-indexcache Oct 10 12:48:27 65K Oct 10 12:48:32 that's nothing really Oct 10 12:48:40 mount should give you all the mounts Oct 10 12:48:48 check the ones with tmpfs ramfs Oct 10 12:49:45 <_trine> https://paste.debian.net/315245/ Oct 10 12:50:49 du -hs /tmp ? Oct 10 12:51:11 <_trine> root@OpenWrt:/tmp# du -hs /tmp Oct 10 12:51:11 <_trine> 24.7M /tmp Oct 10 12:51:11 <_trine> root@OpenWrt:/tmp# Oct 10 12:51:18 free ? Oct 10 12:51:44 <_trine> https://paste.debian.net/315246/ Oct 10 12:52:25 _trine: you see you have half of your memory occupied by something in tmp?= Oct 10 12:54:00 <_trine> yes its luci-modulecache Oct 10 12:54:36 you can go and keep finding out where the rest is, and how much does transmission need etc. Oct 10 12:54:52 <_trine> https://paste.debian.net/315247/ Oct 10 12:54:53 (I don't know if you are using transmission really, but that's the way to debug that) Oct 10 12:55:50 <_trine> I am not running transmission on my router I am running transmission on a dockstar hanging off the router Oct 10 12:56:42 your initial statement was 'my router runs out of memory while downloading torrents but the memory never seems to get reclaimed later' Oct 10 12:56:56 <_trine> exactly Oct 10 12:57:57 <_trine> my router is running out of memory while I am downloading torrents but the torrents are being downloaded on my dockstar not the router Oct 10 12:58:21 <_trine> nevertheless the memory on the router gets used up Oct 10 12:59:30 <_trine> the conntrackd-stats.log is 24,010KiB Oct 10 12:59:59 <_trine> but I dont understand why the memory is not reclaimed after the torrents have finished Oct 10 13:02:43 <_trine> and i dont understand what has happened to it either Oct 10 13:07:07 <_trine> this is from top Oct 10 13:07:10 <_trine> Mem: 58700K used, 2108K free, 24028K shrd, 1468K buff, 28672K cached Oct 10 13:07:10 <_trine> CPU: 0% usr 0% sys 0% nic 99% idle 0% io 0% irq 0% sirq Oct 10 13:07:11 <_trine> Load average: 0.00 0.01 0.05 1/61 21057 Oct 10 13:10:07 <_trine> why should so much memory be used in the router when the torrents are being downloaded on another device Oct 10 13:12:52 <_trine> I dont understand why downloading torrents on a different device which is connected to the LAN of my router should be able to virtually kill the router Oct 10 13:49:59 conntrack table using too much ram Oct 10 13:50:42 <_trine> yes it looks like it Oct 10 13:50:56 <_trine> but what to do,, that is the question Oct 10 13:51:00 i thought that was a thing of the past tho - and only happening on old factory firmware Oct 10 13:51:25 <_trine> maybe Lazarus lives Oct 10 13:52:05 <_trine> this is a fairly new trunk Oct 10 13:52:28 <_trine> on the router and an older one on the Dockstar where the torrents are running Oct 10 13:52:56 cat /proc/net/nf_conntrack Oct 10 13:53:15 an look how many lines there are/entries on torrent Oct 10 13:54:42 <_trine> the torrents have finished now Oct 10 13:55:00 <_trine> I could start a couple though Oct 10 13:55:21 limiting the config of transmission to something reasonable might help too Oct 10 13:55:56 <_trine> whats baffleing me is that its the routers RAM thats getting hammered and its not running the torrents Oct 10 13:57:37 well that happened in the past - but now most torrent clients are reasonably configured regarding nr of connections... and openwrt normally has sane timeouts for udp +tcp conntrack Oct 10 14:02:00 _trine: that number seems normal Oct 10 14:02:26 <_trine> well then I still dont understand this Oct 10 14:03:56 <_trine> "download-queue-size": 3, << so this is quite conservative Oct 10 14:04:35 on router do cat /proc/meminfo - and check those values - there is probably some documentation which values are important Oct 10 14:07:38 <_trine> looks reasonable https://paste.debian.net/315251/ Oct 10 14:09:05 <_trine> this is a wndr3700 with 32meg of ram Oct 10 14:10:03 wndr3700 should have 64mb ram Oct 10 14:10:26 <_trine> err maybe I made a mistake yes Oct 10 14:11:32 <_trine> total used free shared buffers Oct 10 14:11:33 <_trine> Mem: 60808 52576 8232 18856 524 Oct 10 14:12:18 https://paste.debian.net/315252/ thats from one wndr3700 with some older trunk / 3.18 kernel Oct 10 14:12:20 <_trine> the Dockstar has 128m Oct 10 14:13:11 that memory is with 20days uptime and in a lan that has some torrent client too (on a synology tho) Oct 10 14:14:34 oh nice 4500 conntrack entries there Oct 10 14:17:55 _trine: active/inactive(anon) entries are strange Oct 10 14:18:17 <_trine> it seems like the ram on the router is being used up and hardly anything in the dockstar Oct 10 14:18:36 _trine: and shmem on your router is 20M compared to mine which uses like 760k Oct 10 14:18:55 <_trine> can I alter that ? Oct 10 14:19:12 <_trine> I wonder what is causing it Oct 10 14:27:46 maybe its some process running - dont know - the documentation for current kernel is http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/Documentation/filesystems/proc.txt Oct 10 14:29:59 rebooting the router probably should fix the memory issue for a short time and then you could try to check regularly with some cron job how much memory you are using over time and *wait* until your bug shows up again Oct 10 14:30:19 <_trine> rebooting does fix it Oct 10 14:31:39 so observing proc/meminfo and send the info to some network logfile on smb or so and look at the initial values of the velues that are off Oct 10 14:31:53 <_trine> what seems to cause the problem is when I download a torrent on the dockstar Oct 10 14:32:11 <_trine> and the DS seems to be unaffected Oct 10 14:32:39 at least replicating the problem isnt taking up much time ^^ Oct 10 14:33:06 <_trine> I have compiled and flashed several different trunks recently but they have all been the same Oct 10 14:33:57 <_trine> sure I can replicated easily enough Oct 10 14:34:48 <_trine> it ^^ Oct 10 14:36:01 <_trine> sorry dyslexic fingers again Oct 10 14:36:17 <_trine> sure it can be replicated easily enough Oct 10 14:38:48 build #100 of avr32 is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/100 Oct 10 14:46:58 build #95 of ep93xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/95 Oct 10 14:47:23 build #96 of adm8668 is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/adm8668/builds/96 Oct 10 14:47:23 build #96 of uml is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/96 Oct 10 14:47:29 build #95 of mvebu is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/mvebu/builds/95 Oct 10 14:58:26 how can I map the svn revision to the git id? Oct 10 14:59:19 git log -1 --pretty=%H --grep=@12345 Oct 10 15:00:10 thanks! Oct 10 15:14:02 build #108 of ramips.mt7628 is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.mt7628/builds/108 Oct 10 15:14:34 build #97 of ath25 is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/ath25/builds/97 Oct 10 15:14:34 build #95 of mpc85xx is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/mpc85xx/builds/95 Oct 10 15:18:35 build #109 of brcm47xx.legacy is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx.legacy/builds/109 Oct 10 15:19:10 build #94 of mcs814x is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/mcs814x/builds/94 Oct 10 15:21:30 build #108 of cobalt is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/108 Oct 10 15:21:59 build #108 of cns21xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/cns21xx/builds/108 Oct 10 15:24:43 build #108 of orion is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/108 Oct 10 16:58:57 build #102 of brcm2708 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/brcm2708/builds/102 Oct 10 17:09:25 build #105 of ar71xx.nand is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/ar71xx.nand/builds/105 Oct 10 17:36:21 <_trine> plntyk, after rebooting I have got rid of that 20meg of shared memory https://paste.debian.net/315271/ Oct 10 17:39:03 did you have a 20meg log file from contrack? Oct 10 17:39:10 if that's on /tmp that's taking up ram too Oct 10 17:40:18 jow_laptop: https://github.com/openwrt/luci/issues/511 Oct 10 17:40:23 <_trine> karlp, yes Oct 10 17:40:34 turn off the logging? Oct 10 17:40:40 <_trine> how Oct 10 17:59:39 <_trine> in /etc/conntrackd/conntracked.conf it says under Stats # If you enable this option, the daemon writes the information about # destroyed connections to a logfile. Default is off. # Logfile: on, off, or a filename # Default file: (/var/log/conntrackd-stats.log) Oct 10 18:00:34 <_trine> but on my router is was set to 'LogFile on' Oct 10 18:01:01 <_trine> from the default condition Oct 10 18:03:22 well, http://git.openwrt.org/?p=openwrt.git;a=blob;f=package/network/utils/conntrack-tools/Makefile;h=978467b01657ea21c72e0fe1e6437432c1155e3e;hb=HEAD and the init script don't turn it on, Oct 10 18:04:39 <_trine> from what I can see it should have been off but it was on Oct 10 18:05:43 I can't find the conntrackd.conf file in the conntrack-tools package, Oct 10 18:06:40 well, turn it off unless you want it logged. (you may hve been debugging somethign earlier and turned it on?) Oct 10 18:06:58 <_trine> no I have not turned it on Oct 10 18:07:23 <_trine> maybe there is a hash missing from in front of it Oct 10 18:07:42 well, you can either turn it off and move on, or you can rebuild the package and see where the config file comes in as. Oct 10 18:07:59 but if you're ending up with multimeg log files in /tmp, then that's where your ram is going. Oct 10 18:08:14 <_trine> https://paste.debian.net/315275/ Oct 10 18:08:36 <_trine> I have set it to off now Oct 10 18:09:15 <_trine> # Default file: (/var/log/conntrackd-stats.log) Oct 10 18:09:15 <_trine> # Oct 10 18:09:15 <_trine> LogFile off Oct 10 18:13:53 <_trine> I think I'll give some torrents some welly now to test it Oct 10 18:14:23 <_trine> avast you pirates Oct 10 18:20:45 <_trine> perfect Oct 10 18:21:02 <_trine> jow_laptop, you need to take a look at that file please Oct 10 18:25:05 not much to look at, patch the conf to off for stats Oct 10 18:25:26 <_trine> swalker, yes but it needs doing Oct 10 18:26:35 <_trine> swalker, although it should have been set to off as the default it was actually set to on Oct 10 18:26:45 <_trine> and was eating up all my ram Oct 10 18:27:01 you're the man to do it... Oct 10 18:27:24 right, the conf is untouched from upstream Oct 10 18:27:26 <_trine> what can I do I'm just a lowly surf Oct 10 18:27:28 <_trine> :) Oct 10 18:28:07 * swalker would just remove the package from the box unless there's an actual use case Oct 10 18:28:38 <_trine> well setting it to off cured my problem Oct 10 18:33:09 <_trine> swalker, you're right I will compile again without it Oct 10 18:39:19 <_trine> swalker, is it still a bug for upstream seeing as it says the default should be off when it is actually on ? Oct 10 18:45:35 <_trine> network/firewall conntrack-tools was the culprit Oct 10 18:47:47 _trine: maybe, but you're looking at http://git.netfilter.org/conntrack-tools/tree/doc/stats/conntrackd.conf#n37 rather than http://git.netfilter.org/conntrack-tools/tree/doc/stats/conntrackd.conf#n125 Oct 10 18:52:11 <_trine> swalker, where does line 37 come from Oct 10 18:52:21 <_trine> I was looking at line 125 Oct 10 19:01:30 I'd file it upstream, it seems unusual for one log to default to off and one to default to on. Oct 10 19:03:22 <_trine> karlp, it does not default to on I think someone has mised out a # in front of the line Oct 10 19:04:06 only been there for about 7 yrs Oct 10 19:04:35 <_trine> lol trine bug finder general Oct 10 19:04:46 _trine: well, maybe they didn't _intend_ to, but it _is_ default on. Oct 10 19:05:23 <_trine> well at least its been found now Oct 10 19:06:48 meaningless unless you file itupstream :) Oct 10 19:08:23 <_trine> swalker, would you be so kind as to file that bug up stream for me please Oct 10 19:16:02 someone running luci on nginx? or lighttp? Oct 10 20:04:17 mhei does iirc Oct 10 20:51:33 build #101 of x86.kvm_guest is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/x86.kvm_guest/builds/101 Oct 10 21:50:22 build #107 of pxa is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/pxa/builds/107 Oct 10 22:23:12 build #108 of realview is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/realview/builds/108 Oct 10 22:27:06 build #108 of rb532 is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/108 Oct 10 22:27:52 build #108 of imx6 is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/imx6/builds/108 Oct 10 22:28:20 build #107 of x86.64 is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/x86.64/builds/107 Oct 10 23:25:15 build #99 of ar71xx is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/ar71xx/builds/99 Oct 11 02:03:02 build #109 of lantiq is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/lantiq/builds/109 Oct 11 02:37:50 am I still the only one that gets all flashing lights on archer c7 when flashing it? Oct 11 02:38:28 I can't seem to build a working firmware even though my changes are super minimal like adding full dnsmasq instead of minimal Oct 11 02:38:33 image is plenty small, etc Oct 11 02:44:03 [mktplinkfw] kernel length aligned to 1042320 Oct 11 02:44:03 [mktplinkfw] *** error: images are too big Oct 11 02:44:15 ahh, something is wrong but it is non-fatal Oct 11 02:44:19 that's pretty annoying Oct 11 02:45:00 nbd: forgive me, I don't really know devs other than yourself. I am hitting an error that I think should be fatal and it isn't. ping me when you are around please Oct 11 02:50:10 nbd: I notice two errors, one, it fails to build images for c7v1 because my image is too big (I have v2 so I don't care) Oct 11 02:50:21 nbd: but the other error seems bad: Makefile:2105: recipe for target 'image_prepare' failed Oct 11 02:50:51 nbd: seems to fail because of this: cp: '/home/zero/development/openwrt/build_dir/target-mips_34kc_uClibc-0.9.33.2/linux-ar71xx_generic/loader-generic.elf' and '/home/zero/development/openwrt/build_dir/target-mips_34kc_uClibc-0.9.33.2/linux-ar71xx_generic/loader-generic.elf' are the same file Oct 11 02:53:16 build #96 of mpc83xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/mpc83xx/builds/96 Oct 11 02:56:17 Zero_Chaos: ./scripts/diffconfig.sh will point out the number of changes Oct 11 02:56:51 haven't seen the latter one myself Oct 11 02:57:36 swalker: I honestly can't imagine why that would cause a failure, but the fact that it is copying a file into itself seems to suggest something else going wrong **** ENDING LOGGING AT Sun Oct 11 02:59:58 2015