**** BEGIN LOGGING AT Sun Jan 24 02:59:57 2021 Jan 24 04:45:16 mangix: test Jan 24 04:45:48 nyt: what's the conclusion with x86 metadata? Jan 24 05:02:52 aparcar[m]: I guess https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=29167cbca3653de05a8b915bc21327dac7d05174 is supposed to sort it Jan 24 05:12:02 pkgadd: thanks, I had the same idea happy to see it's fixed Jan 24 07:06:36 aparcar[m]: switched from irssi to quassl Jan 24 07:06:44 seems to work so far Jan 24 08:40:54 mangix: quassel will work. see no reason why it won't. i am using exclusively quassel-core with postgresql backend on one of the arm nodes and one of the quassel clients on devices like tablets, phones or laptops Jan 24 08:44:00 Yeah. Looks like on mobile I have to use quassel-web. The Android app works poorly it seems. Jan 24 08:44:24 damex: Jan 24 08:44:36 mangix: which android app ? :) Jan 24 08:44:57 https://quasseldroid.info this one ? Jan 24 08:45:26 Yeah Jan 24 08:52:18 mangix: should work pretty well. i would suggest contacting them at #quasseldroid - it is one of the best mobile clients out there Jan 24 08:52:59 currently i am on iOS so using quassel-web or woboquassel (worse experience compared to a web version) Jan 24 09:00:19 I'm using weechat (I should try to packaged it to OpenWrt) with Glowing Bear or just use quassel as you suggest as it is in OpenWrt. Jan 24 09:04:25 Pepe: I run quassel in a docker instance on a NAS Jan 24 09:06:25 damex: no idea about iOS Jan 24 09:08:35 speaking of docker, I just redid my setup. It sucks running services on a low powered ARM device Jan 24 09:55:13 mangix: it gets better when there is Jan 24 12:20:55 seems bit of a pity that umurmur has not been updated for 15 months, must be some reason for that i guess Jan 24 13:52:00 Hey all :) Jan 24 18:11:03 updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html Jan 24 18:15:07 grift: There were two releases of umrumur in 21 days. See https://github.com/umurmur/umurmur/releases and in OpenWrt packages feed pull request with bumping a version will lend soon Jan 24 18:16:47 nice thanks, yes i noticed that upstream Jan 24 18:17:34 ill look into adding support for it in the selinux security policy once a new package drops, its a neat poor mans voip solution Jan 24 18:39:38 how can I get rid of the annoying feeds install -a warning: "WARNING: Not overriding core package 'iperf'; use -f to force" Jan 24 18:40:18 './scripts/feeds install -a -f' will still report: "Overriding core package 'iperf3' with version from packages" Jan 24 18:40:45 how can I flush this away, do I need manually uninstall them from core? Jan 24 18:40:59 rm -rf tmp (NOT rm -rf /tmp) Jan 24 18:41:09 everything a package is removed from core this came up... Jan 24 18:42:24 ldir: thanks! will this cause a longer 'make' rebuild as it has to re-assembly the tmp dir? Jan 24 18:42:53 s/everything/everytime/ Jan 24 18:43:20 it's a little longer as it re-finds all the packages but not significant Jan 24 18:43:51 it shouldn't prompt a rebuild of the entire toolchain if that's what you're worried about! Jan 24 18:47:29 * ldir sets off to drive home in the snow - hope I make it Jan 24 18:52:25 safe driving, thanks! Jan 24 18:55:02 ynezz: did you get any testing confirmations concerning 1fb413e6579a34a0040b526e681298909dfaa5ac ? Jan 24 19:00:27 just asking because i need to open up my device for serial, in case there have been no testing reports. Jan 24 19:12:21 I decided I was going to repurpose a spare laptop into a WSL2 box to build with :) So far, it's been very smooth Jan 24 19:13:38 Grommish: why not linux? Jan 24 19:14:27 Borromini: I do have a standard Ubuntu server box I use, but I wanted to be able to test Jan 24 19:14:47 and O"ve never used WSL before :) Jan 24 19:15:24 =) Jan 24 19:15:36 Honestly, it's exactly the same with the single step of removing Windows paths Jan 24 19:15:45 Even Rust is building :D Jan 24 19:16:20 isn't it a full vm with wsl2? Jan 24 19:16:54 I was running Server2016 on the box to act as a WSUS and PXE boot server and decided it was too much of a PITA to maintain Jan 24 19:17:01 Borromini as far as I can tell yes Jan 24 19:19:17 but, the i7-7500u is faster than the i7-37xx in the other box, even if it's VMed.. I don't have quite as much RAM in this box, but I think it should e quicker to build this way Jan 24 19:19:42 time for zen 3 eh ;) Jan 24 19:19:44 * Borromini hides Jan 24 19:19:56 I looked at a Dual Epyc Rome board Jan 24 19:20:06 but NewEgg wants 14k for the Mobo/CPU Jan 24 19:20:15 I consider that tasteful Tech p0rn Jan 24 19:20:36 but a legit -j256 would be yummy Jan 24 19:20:54 and obscene, but fun Jan 24 19:21:42 Grommish: IMHO: as long you don't build everything, for a basic openwrt images you don't rellay need more than -j 18 Jan 24 19:21:50 https://www.newegg.com/supermicro-mbd-h11dsi-n702-ma015-o-dual-amd-epyc-7000-series/p/N82E16813183691 Jan 24 19:22:13 I just -j9 now and it's enough, and I do build CONFIG_ALL from time to time, but only for single targets Jan 24 19:23:18 * Borromini is on -j16 but zen 1 Jan 24 19:23:52 But again, I just looking at it :) I can't drop that on a server box.. and it doesn't even come with a case or others. like the PSU to run it haha That'll be easiler 20k+ by the time you're done with it Jan 24 19:24:15 I'd have to get another rack :x Jan 24 19:24:50 Also, isn't it the general consensus -j should be nproc*2+1? Jan 24 19:25:10 Or I guess Thread*2+1 Jan 24 19:25:14 :D Jan 24 19:25:38 I was on the WSL2 wiki and there was general conflicting messages and confusion Jan 24 19:25:44 but I've always used the thread*2+1 Jan 24 19:25:44 Grommish: i thought so too, until i bumped into why the actual amount of threads (and not +1) is better Jan 24 19:25:54 Hmm Jan 24 19:26:10 I'll have to check a timed build.. I'd be interested to see.. Jan 24 19:26:10 1800X 8c16t here and i used to do -j17 Jan 24 19:27:11 is it an overhead issue? Jan 24 19:29:41 no idea. Jan 24 19:30:54 All of my baseline times are skewed because it's based on the other machine.. slower Cpu, more RAM and an SSD rather than a HDD.. I wonder how badly that is going to tank, or if it's all mostly memory bound Jan 24 19:33:57 i did a few runs between an S-ATA and NVMe SSD a while ago Jan 24 19:34:01 barely any difference, really Jan 24 19:34:38 I do -j15 on my ryzen 7 2700 and after looking at the cpu usage through out the build there is quite a lot of time when the CPU is wateing for data from the drive or the memmory. Jan 24 19:34:47 Grommish: but that won't help you since you're talking HDD and SSD, sorry Jan 24 19:35:39 I don't see much of a diff from a sata 3 ssd or a m.2 drive. Jan 24 19:35:54 nope Jan 24 19:36:15 I had to decide between a 480SSD or a 2TB 7200 hehe Jan 24 19:36:20 I had to cop to the space Jan 24 19:37:00 I have 16 gig memmory and a build including building the toolchane will use up to 11.75 gig Jan 24 19:37:31 Grommish: storage is HDD here as well. just not on the desktop, but in a home server Jan 24 19:37:54 I can see why you made that chose Grommish Jan 24 19:38:06 Borromini, what the build time on that? Jan 24 19:38:08 *nod* I've got 24GB Ram in this laptop, and it's at least using it Jan 24 19:38:20 like, from clean master minus download times for deps Jan 24 19:38:29 That most be some laptop! Jan 24 19:38:41 must* Jan 24 19:38:42 shibboleth: i gotta check sometime, with or without toolchain? Jan 24 19:39:00 i dumped ccache a while ago so times have gone up :P Jan 24 19:39:27 ccache seems to mess up to mutch. Jan 24 19:39:33 I've got an AW-M17xR4 for one machine.. that has the i7-37xx, 32GB ram, 64GB m.2 boot, 500GB SSD and 1TB HHD for storage.. this one is the i7-7500u, 24GB ram, and the HHD Jan 24 19:39:40 git clone master, scripts/feeds update/install, make defconfig, make mencuonfig, c2600, make -j16 Jan 24 19:39:48 ccache was more trouble than it's worth Jan 24 19:39:59 * Tapper nods! Jan 24 19:41:51 shibboleth: I'll run a c2600 run minus the rust stuff to give ya a time if you want.. i need to do it anyway. and it'l require a toolchain compile Jan 24 19:42:01 nice Jan 24 19:42:11 I just need to finish this build Jan 24 19:42:27 i have an xeon 1235v3 doing it in about 20 minutes Jan 24 19:42:49 With the toolchain I'm around 45-60 I think on the othe rmachine Jan 24 19:43:03 from scratch, few minutes for secondaries Jan 24 19:43:23 I had a couple of iDataPlex 645s a while back for building.. ended up having to get rid of them:( Jan 24 19:43:53 shibboleth: is c2600 ipq806x? Jan 24 19:44:01 yes Jan 24 19:44:18 k Jan 24 19:44:19 like, keep+ the dl folder ofc Jan 24 19:44:22 i'll give it a shot as well Jan 24 19:44:49 shibbolet Any chance you'd want to do me a favor and test a package if you get bored and time? :D Jan 24 19:44:52 just the make though (but will be a new toolchain, i don't have any ipq806x) Jan 24 19:45:21 Grommish, "test a package"? Jan 24 19:45:27 :D Jan 24 19:45:52 thought i'd throw a dual-socket x5690 at building openwrt just for shits and giggles, but ofc it's gonna suck Jan 24 19:46:37 i'm thinking the 1235 is gonna crush it, so i haven't bothered Jan 24 19:46:55 https://github.com/openwrt/packages/pull/13916 Jan 24 19:47:09 I need more testing for failure with the rust toolchain Jan 24 19:47:25 because I need to patch the source for each target type Jan 24 19:49:49 Maybe you should first fix the errors according to Github Actions to know more. ;-) Jan 24 19:50:02 I don't want Actions to run at this point Jan 24 19:50:13 Because it's a massive compile for no gain Jan 24 19:50:21 It's a host only package Jan 24 19:51:26 But you can try it to include it with Suricata commit and by using CI you know if it was compile tested and download artifacts which you can use on routers Jan 24 19:51:44 Right Jan 24 19:52:22 and right now even host package is not compiled Jan 24 19:52:24 the target triples uses OpenWrt's default, which is x64_86-openwrt-linux-musl, etc Jan 24 19:52:35 It compiles locally on a fresh system install Jan 24 19:52:48 with the exception of libunwind I had to change to recognize mips64 hehe Jan 24 19:53:11 It's still very much a WIP though Jan 24 19:53:42 I need a way to create/verify the sha256 hash on the created installation binary tarball Jan 24 19:53:43 Package hash check failed in the PR which you mentioned and it is really trivial to fix that Jan 24 19:53:45 WARNING: rust-1.49.0-x86_64-unknown-linux-gnu_powerpc-openwrt-linux-musl-install.tar.xz is missing, please run make download before re-running this check Jan 24 19:53:47 WARNING: PKG_MIRROR_HASH is missing, set to e7c3fefe77ee8c15b49fac4c89c9f7025c57e8f6ff3079d905e51a96bda3042d Jan 24 19:54:26 powerpc-openwrt-linux-musl that is what i need :) Jan 24 19:55:54 PKG_MIRROR_HASH shouldn't be set at all Jan 24 19:56:20 I don't use it because there is no source tarball to get.. Jan 24 19:56:59 But, I have been working on clean up the extranious crap to make it cleaner Jan 24 19:58:29 Then something with your Makefile is wrong, but you should really check CI which is used in package repository ( https://github.com/openwrt/packages/actions/runs/478343032 ) to know more as I said in comments. What's compiled, its uploaded to artefacts which you can download. You can find there logs, packages and the later one comes handy. Also, compile tests will fail because there is PKG_HASH:=skip Jan 24 19:58:31 and there is no chance to pass it with it. You should put there correct has Jan 24 19:58:33 h Jan 24 20:00:08 Hi. I'm running 19.07.5 on MikroTik RouterBOARD 750Gr3 / MediaTek MT7621 ver:1 eco:3 and OpenWrt was dying today after only 11 days of uptime. I have remote syslog but there no messages at all visible. It looks like the IP stack was suddenly dying. When looking at the physical device its LAN LED was still blinking. How can I prepare for a better diagnosis for the next time, idK write syslog on an Jan 24 20:00:14 external USB or flash drive? Please suggest. Jan 24 20:01:07 looks like this issue? https://bugs.openwrt.org/index.php?do=details&task_id=2628 Jan 24 20:02:52 probably / can be Jan 24 20:04:47 Pepe: yes, I need to put the hash in, but I was waiting until I figure out what revision I end up using, but I'll toss it in there Jan 24 20:05:16 shibboleth: https://paste.debian.net/1182591/ build started at 20:49:38 and ended at 21:03:42 Jan 24 20:05:34 this is just the make -j16 though (but with the whole toolchain, so clean compile) Jan 24 20:06:25 mangix: could you please give https://github.com/openwrt/packages/pull/13534 another look? Jan 24 20:06:35 thanks. and this is an amd what? specs Jan 24 20:07:48 1800X at stock speeds, Samsung 970 Evo 1TB NMVe, 32 GB DDR4 2933 Jan 24 20:10:42 22m31s on an ivy/1235v3, 960 sata, 32gb 12800e Jan 24 20:10:51 -j8 Jan 24 20:13:52 now, if anyone has a mac mini/mbp a1 and care to give it a spin? Jan 24 20:14:58 reiffert: take a look if the kernel created a crash. /sys/kernel/debug/crashlog should exists then. Jan 24 20:16:20 sorry, 860 sata Jan 24 20:18:44 lynxis: I was restarting the device two times since the event and there is no file Jan 24 22:48:27 reiffert: this issue is solved in master/snapshot builds Jan 24 22:48:39 there is no easy way to fix for 19.07 Jan 24 22:50:40 aparcar[m]: uhhh why do I remember an OpenSSL patch that I submitted upstream. I can't find it. Jan 24 22:55:33 oh. I sent a merge request to my own forked repo. Worthless. Jan 24 22:57:06 and making a merge request against the source repo was actually difficult Jan 24 22:57:20 What kind of alternative to github is this? Jan 24 22:58:24 aparcar[m]: https://gitlab.alpinelinux.org/alpine/apk-tools/-/merge_requests/53 Jan 24 23:01:59 mangix: thanks, where did this error strike you? it seem to compile fine with the SDKs? Jan 24 23:04:45 aparcar[m]: disable deprecated APIs under libssl Jan 24 23:15:32 mangix: ok thanks Jan 24 23:31:49 aparcar[m]: I think every package is fixed to compile actually. Would be interesting to run a buildbot to see if any package remains Jan 24 23:38:15 mangix: I think a buildbot runs once a day? Jan 25 00:03:41 aparcar[m]: right but deprecated APIs are enabled by default Jan 25 01:23:04 mangix: do you recall the git commit hash? Jan 25 01:24:16 mangix: assuming it was fixed in 5.4 and offloading is gone? Jan 25 01:25:35 reiffert: you can't simply backport it to 19.07 Jan 25 01:25:58 the ethernet driver changed with the update to kernel 5.4 Jan 25 01:29:16 mangix: umdns? I don't know anything aboutit Jan 25 01:29:59 https://patchwork.ozlabs.org/project/openwrt/patch/20201015061027.7614-1-rosenp@gmail.com/ Jan 25 01:30:09 i've been posting it for quite a while... Jan 25 01:31:28 grep -c 2020101506 "#openwrt-devel.log" Jan 25 01:31:28 3 Jan 25 02:52:55 mangix: I'm sorry. I don't use that or at least don't know how to test it **** ENDING LOGGING AT Mon Jan 25 02:59:57 2021