**** BEGIN LOGGING AT Sun Jul 19 02:59:58 2015 Jul 19 05:52:03 i'm trying to learn what sysuprade does by reading the script source, but there is this cmd "ubus call system upgrade", i wonder it does? if I understand ubus correctly, some process has to listen to this upgrade event to respond to this cmd? what this process might that be? what it does? Jul 19 05:54:24 how often is the git repo syncd? Jul 19 06:28:21 nyt: which git repo you mean ? Jul 19 06:28:54 the github openwrt-mirror, nm tho, i had an issue with my local copy Jul 19 06:28:58 trac/svn / git.openwrt.org / nbd.name mirror should be in sync but sometimes trac seem to lag Jul 19 07:39:01 does "mtd write firmware" also wipes out jffs2 partition? from the cmd it seems it only overwrites the firmware partition Jul 19 09:11:14 Hi. A question regarding maintaining a package in the "packages" feed. I'm the maintainer of OpenOCD there, but I just do not understand if and how am I supposed to receive any notifications of _related_ "github pull requests". What are my responsibilities as a package maintainer? How this whole system works? Jul 19 09:19:51 I've read CONTRIBUTING.md before asking, of course. Jul 19 09:23:07 PaulFertser: you probably will only receive a notification if someone assigns the pullrequest to you, or mentions you in a comment (I think github can do that) Jul 19 09:23:53 KanjiMonster: but CONTRIBUTING.md doesn't mandate that from those creating "github pull requests", hence my question. Jul 19 09:26:38 PaulFertser: afaict some of the people with commit rights do add the mentions/assigns for pull request, so ideally you should get one eventually, if you have notifications enabled Jul 19 09:28:43 KanjiMonster: aha, so basically it's the responsibility of those people with commit rights to draw maintainers' attention to any relevant "github pull request"? Well, if they're aware of that, I should be fine just monitoring my email then, and not worrying about anything, right? Same when creating "github pull requests" for other packages, I just create it the usual way, and then wait for the Jul 19 09:28:49 others to notify the maintainer. Do I understand it right? Jul 19 09:29:37 well, if you can, do assign it to the appropriate maintainer Jul 19 09:29:47 or mention him with @ Jul 19 09:30:55 KanjiMonster: should I add this to CONTRIBUTING.md as well? Jul 19 09:31:00 I mean what you just told me. Jul 19 09:32:25 currently thinking about it, it might make sense to have a list of github accounts of the package maintaners to make that easy Jul 19 09:32:46 Exactly my thought. Jul 19 09:33:25 I don't want to enforce something that requires a bit of not-easily done research Jul 19 09:39:50 PaulFertser: look like you can't set github to do automatic assignments (which would have been nice workaround) Jul 19 09:45:52 KanjiMonster: if people wouldn't be allowed to assign others as package maintainers, something like this would work: git show $(git rev-list $(git blame utils/openocd/Makefile | grep PKG_MAINTAINER | cut -f1 -d\ )^..master --ancestry-path --merges | tail -n1) | sed -n '/pull request/ { s^.*from \(.*\)/.*^\1^; p; q }' Jul 19 09:47:52 PaulFertser: but only for those that don't have commit rights Jul 19 09:48:58 KanjiMonster: yes :/ still, I hope you can see why I'm not happy about the current situation. That github is such a pain in every regard, I just can't understand how people are supposed to use it comfortably. Jul 19 09:52:20 Probably github api search can be used here? Jul 19 09:55:05 KanjiMonster: wget -q -O- 'https://api.github.com/search/users?q=fercerpav@gmail.com' | sed -n '/login/ { s^.* "\(.*\)",^\1^; p; q }' Jul 19 09:56:36 Doesn't work for users using one email for github and another for PKG_MAINTAINER Jul 19 10:01:39 PaulFertser: I guess it's best to create a list of package maintainer accounts (probably in the wiki), so you can also use it for assigning issues as well Jul 19 10:03:30 A manually maintained one?.. Jul 19 10:10:16 They say one can use https://developer.github.com/webhooks/ to run an external script whenever a "github pull request" is created. Jul 19 10:23:38 PaulFertser: AFAICT, this would require an external server to run the script Jul 19 10:24:11 KanjiMonster: yes :/ Jul 19 11:42:42 KanjiMonster: i'm noticing some stalls on my mikrotik rb493g with r46176. i'm wondering if it might be related to the interrupt disabling to fix the switch problems earlier. Jul 19 11:43:08 russell--: stalls? Jul 19 11:43:42 like, ath5k radio stops passing traffic Jul 19 11:44:24 my client device, e.g. this laptop i'm typing on right now, shows it is associated with a nice signal, but i can't ping my gateway Jul 19 11:46:32 intermittently Jul 19 11:47:26 works, then doesn't work for ~30-60 seconds, then works again. nothing too suspicious shows up in the syslog or dmesg. Jul 19 11:47:56 hm, it shouldn't cause 30~60s delays Jul 19 11:48:56 I do see these in dmesg and on the console: [1257935.650000] sit: Src spoofed 205.171.2.64/2602::205:171:3:250 -> 63.228.172.191/2602:3f:e4ac:bf01:2916:a38e:926c:109b Jul 19 11:49:25 russell--: you should be able to test this by not using the eth-whatever-it-was at all in your config so it is down (then the mdio bus won't be accessed), and checking if it still happens Jul 19 11:49:50 okay, will do Jul 19 11:57:12 i was seeing 4% packet loss in my laptop to gateway ping (over about 600-700 pings), i took eth0 out of the bridge and shows down, retesting. Jul 19 12:00:01 hmm, on 300 pings with eth0 down, still seeing 4% packet loss, looks like something else then. Jul 19 12:02:11 the stalls are probably just due to the packet loss Jul 19 12:11:48 changed channels and packet loss goes to 1 packet in 300 ( <1%) Jul 19 12:12:09 so, looks like wireless interference is all Jul 19 12:50:22 i'm trying to improve the sysupgrade script by making it wipe out the overlay even if it's from extroot, i wonder will simply removing the upper dir in it be enough? Jul 19 13:01:01 hgl: probably; that's what the firstboot script does when rootfs_data is mounted; but please make it non-default behaviour; I would assume many use extroot as more than just a larger overlay, but also to store userdata and it suddenly getting wiped on sysupgrade might be a bit ugh surprising Jul 19 13:01:44 e.g. I have home directories for samba on my overlay Jul 19 13:02:52 but if you don't use extroot, the overlay is wiped anyway, i think wiping out extroot is in consistent with the default behavior? Jul 19 13:03:57 I think you should put data outside of the upper folder and then symlink to it, so they can survive the upgrade Jul 19 13:13:14 i'm also thinking about auto installing usb related packages to mtd after the upgrade (also the corresponding fs package for the extroot fs), and only backing up /etc/config/fstab to mtd, with other configs saved to extroot, so extroot is automatically loaded after restarting, in which case the overlay must be wiped out to prevent bricking the device. I wonder if this automation sounds good? Jul 19 13:35:27 hgl: currently it isn't wiping the overlay on overlay, so there *will* be people that rely on it, and wiping it is destructive and irreversible. So it should really not be the default behaviour for extroot Jul 19 13:43:26 KanjiMonster, i see you concern. i think i can make it into a flag, disabled by default, only when enabled, extroot support is added, and thus the usb overlay is wiped out. Jul 19 13:47:42 hgl: a flag to pass to sysupgrade is fine, this can then be easily exposed in luci Jul 19 14:14:01 build #66 of brcm63xx.smp is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/brcm63xx.smp/builds/66 Jul 19 14:20:47 I summarized my plan here, also asked a few questions, would like to get some help, thanks. https://gist.github.com/hgl/f45c4f40e5beec74fe06 Jul 19 14:59:37 nbd r46422 trunk/package/devel/strace/Makefile * strace: cross compile fix: pass HOST_*FLAGS in to ./configure Jul 19 14:59:43 nbd r46423 trunk/package/devel/strace/Makefile Jul 19 14:59:43 strace: add kernel headers include to host cflags to fix build on non-linux systems Jul 19 14:59:53 nbd r46424 trunk/tools/mkimage/Makefile * mkimage: cross compile fix: pass HOST_*FLAGS in to uboot's makefile Jul 19 17:58:53 nbd r46425 trunk/target/linux/ar71xx/ (8 files in 8 dirs) * Revert "ar71xx: Add support for Cisco-Linksys WAP4410N" (r46250) Jul 19 17:59:07 nbd r46426 trunk/target/linux/ar71xx/ (96 files in 2 dirs) * ar71xx: add 4.1 support Jul 19 17:59:13 nbd r46427 trunk/target/linux/ar71xx/ (39 files) * ar71xx: refresh 4.1 patches Jul 19 17:59:20 nbd r46428 trunk/target/linux/generic/config-4.1 * kernel: add missing config symbols for 4.1 Jul 19 17:59:27 nbd r46429 trunk/target/linux/ar71xx/config-4.1 * ar71xx: refresh kernel config Jul 19 17:59:35 nbd r46430 trunk/target/linux/ar71xx/ (50 files) * ar71xx: reorganize 4.1 patch directory layout Jul 19 17:59:42 nbd r46431 trunk/target/linux/ar71xx/patches-4.1/910-unaligned_access_hacks.patch * ar71xx: add missing patch chunk to 4.1 unaligned access hacks Jul 19 17:59:47 nbd r46432 trunk/target/linux/ar71xx/Makefile * ar71xx: switch to linux 4.1 Jul 19 19:37:11 build #63 of mpc83xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/mpc83xx/builds/63 Jul 19 20:13:22 updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html Jul 19 21:16:23 build #63 of ep93xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/63 **** ENDING LOGGING AT Mon Jul 20 02:59:58 2015