| mup | PR snapd#8001 opened: boot: enable UC20 kernel extraction and bootState20 handling <UC20> <Created by anonymouse64> <https://github.com/snapcore/snapd/pull/8001> | 04:53 |
|---|---|---|
| mborzecki | morning | 06:31 |
| mborzecki | interesting postrm test failure on debian: https://paste.ubuntu.com/p/WVGk8pW7rF/ | 06:47 |
| mup | PR snapd#8002 opened: tests/core18/snapd-failover: collect more debug info <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/8002> | 07:08 |
| mborzecki | another one in 2.43.1 PR https://paste.ubuntu.com/p/F2WWvNc3Gv/ | 07:35 |
| mborzecki | mvo: hey | 07:51 |
| mvo | mborzecki: good morning! | 07:52 |
| mborzecki | mvo: can you upload the snapd release tarballs for 43 and 43.1? | 07:53 |
| mvo | mborzecki: yes, sorry, will do that now! | 08:02 |
| mborzecki | mvo: thank you! | 08:02 |
| mvo | mborzecki: also need to update the roadmap and some more things related to the release. there were some build failures (mostly 7997 related) that delayed things | 08:02 |
| mvo | mborzecki: big thanks for 8002 | 08:03 |
| pstolowski | morning | 08:03 |
| mborzecki | pstolowski: hey | 08:03 |
| mvo | hey pstolowski - good morning! | 08:03 |
| pstolowski | o/ | 08:03 |
| mborzecki | mvo: btw. some weird interaction between apt command-not-found and snap advise https://paste.ubuntu.com/p/F2WWvNc3Gv/ | 08:03 |
| mvo | pstolowski: the server team is asking for a meeting about boot speed today, I added you, hope you have time. you will also need to refresh my memories where we stand here | 08:04 |
| pstolowski | mvo: sure, thanks for invite | 08:04 |
| mvo | mborzecki: this logs fails with snapd : Depends: apparmor (>= 2.10.95-0ubuntu2.2) but 2.10.95-0ubuntu2 is to be installed - which looks unrelated to c-n-f. what am I missing :) ? | 08:05 |
| mborzecki | mvo: uh,w8 not that one, this one https://paste.ubuntu.com/p/WVGk8pW7rF/ | 08:05 |
| mvo | mborzecki: looking | 08:09 |
| pstolowski | mvo: it got side-tracked by the need of fixing services (generating wrappers, enabling them with regard to install hook and start-snap-services). first PR of the firsboot stuff landed (snap-preseed command), but will become usable once the snapd changes PR lands (blocked by services). services fixes have one open PR atm, there will be one more touching snapctl side. | 08:09 |
| mvo | mborzecki: thanks, that " E: Could not read response to hello message from hook [ ! -f /usr/bin/snap ] || /usr/bin/snap advise-snap --from-apt 2>/dev/null || true: Broken pipe | 08:10 |
| mvo | quiet: end of output" might be interessting for juliank. I remember him fixing a issue there a while ago but it seems like it's coming back | 08:10 |
| zyga | good morning | 08:11 |
| mvo | pstolowski: that's what I remember, I wonder if we could build a version that works without services or is that just impossible? i.e. we need the changes there | 08:11 |
| mborzecki | mvo: oh cool, i was a bit confused that it'd call cnf handler on removal too | 08:11 |
| mborzecki | zyga: hey | 08:11 |
| zyga | hey, what's up? | 08:12 |
| juliank | mborzecki: what's the apt version? | 08:12 |
| mborzecki | juliank: let me check | 08:12 |
| zyga | I'll try to do coding today but my hand hurts even more :/ | 08:15 |
| zyga | I wonder if this is that wrist injury that some coders have | 08:15 |
| zyga | and that little wrestling with Janek is unrelated | 08:15 |
| zyga | or i can type with just one hand, dunnop | 08:16 |
| mvo | mborzecki: I don't remember the details but it's a bit strange | 08:16 |
| mvo | mborzecki: tarballs updated | 08:23 |
| mborzecki | juliank: it's 1.8.0~alpha3, though i see there's 1.8.4 uploaded already but we don't update it in the tests | 08:24 |
| mborzecki | mvo: thanks! | 08:24 |
| juliank | mborzecki: that needs updating, that's even a pre-release apt | 08:24 |
| mborzecki | juliank: got it, thanks! | 08:25 |
| mvo | mborzecki: sounds like sergio needs to update the images? | 08:26 |
| mborzecki | mvo: yup | 08:26 |
| mborzecki | mvo: i'll ping him when he gets online | 08:26 |
| mvo | juliank: thank you! and sorry for the noise, we should have figured this out ourselfs | 08:26 |
| mvo | mborzecki: thank you | 08:26 |
| juliank | just let me double check | 08:26 |
| juliank | yup | 08:27 |
| juliank | that bug was fixed in 1.8.0~rc1 | 08:27 |
| pedronis | mvo: mborzecki: we need to chat about boot partition layouts, because I probably misunderstood something, and we are producing confusing code as a consequence, also I'm not sure the differences between 20 and 18/16 for boot have a reason or just arbitrary | 08:27 |
| mborzecki | juliank: thanks for double checking! we'll update the spread images, and hopefully the problem will go away ;) | 08:28 |
| mvo | pedronis: sure, when do you want to chat? | 08:30 |
| pstolowski | mvo: yes, it's possible, that's what i did originally | 08:31 |
| pedronis | mvo: mborzecki: in 5-10 mins ? | 08:32 |
| mvo | pstolowski: ok, maybe worth a chat with pedronis if we can give the server team something before we nail services | 08:32 |
| mvo | pedronis: works for me, standup channel in 10min? | 08:33 |
| pedronis | yes | 08:33 |
| mborzecki | pedronis: sure, let me grab some coffee | 08:34 |
| pstolowski | mvo: sure, sounds good | 08:35 |
| pedronis | mborzecki: mvo: I'm in the standup | 08:41 |
| zyga | afk for a moment | 08:59 |
| zyga | re | 09:09 |
| pstolowski | mvo, pedronis let me know if/when we can talk about firstboot & services | 09:12 |
| mvo | pstolowski: either around 11:30 or after the standup, depends how my meeting at 11:00 goes but probably after the standup | 09:18 |
| mvo | pstolowski: does not have to be long, just making sure we have a common understanding going into the other meeting today with server | 09:19 |
| pstolowski | mvo: ok, either sounds fine | 09:20 |
| * zyga goes upstairs to make cofee | 10:04 | |
| zyga | *coffee | 10:04 |
| mup | PR snapd#8003 opened: o/ifacestate, api: implementation of snap disconnect --forget <Created by stolowski> <https://github.com/snapcore/snapd/pull/8003> | 10:18 |
| sdhd-sascha | hi, | 10:18 |
| sdhd-sascha | i have trouble with glibc. Is it possible to deliver my own glibc inside a snap ? | 10:19 |
| zyga | sdhd-sascha: yes though I'd like to understand why you think that is required | 10:19 |
| zyga | I'll be back shortly, I really need to make that coffee | 10:20 |
| * zyga gets up and leaves | 10:20 | |
| sdhd-sascha | zyga: i use `libsnapcraft-preload.so`. And in `canonicalize.c` is a macro: "`#if SHLIB_COMPAT(libc, GLIBC_2_0, GLIBC_2_3)`". I want to remove this part, because the `realpath` function only works if the `resolved` is non NULL | 10:22 |
| sdhd-sascha | zyga: i already asked sergiusens. But he's really busy. | 10:24 |
| sdhd-sascha | Maybe there is another solution, which i didn't see. (My C++ knowledge is very old and non-practical) | 10:24 |
| sdhd-sascha | This is the source, which works only without `snapcraft-preload`. | 10:27 |
| sdhd-sascha | With `snapcraft-preload` one call of `realpath` failed. | 10:27 |
| sdhd-sascha | https://www.irccloud.com/pastebin/wyZvthyv/ | 10:27 |
| sdhd-sascha | Is there already a non-compat glibc which could be packaged with a snap ? | 10:31 |
| zyga | re | 10:32 |
| sdhd-sascha | :) | 10:32 |
| * sdhd-sascha lunch | 10:32 | |
| mborzecki | hmmm - Download snap "strace-static" (18) from channel "stable" (http2: server sent GOAWAY and closed the connection; LastStreamID=1, ErrCode=NO_ERROR, debug="") | 10:42 |
| mvo | GOAWAY is rude | 10:42 |
| * mvo shakes fist | 10:42 | |
| mborzecki | log if anyone's interested: https://paste.ubuntu.com/p/NMRnvTRTqP/ | 10:43 |
| zyga | mvo: we should rename it to PLEASESTOP | 10:43 |
| mborzecki | zyga: mvo: and the client sends BENICE | 10:43 |
| zyga | it should just return BEHAVE! | 10:43 |
| mvo | :-D | 10:43 |
| mborzecki | btw https://github.com/systemd/systemd/issues/10872 is fixed, i'm trying the reproducer right now to confirm it's fine | 10:46 |
| mborzecki | (across a decent number of runs) | 10:46 |
| Chipaca | is http2 secretly intercal | 10:48 |
| pedronis | mvo: mborzecki: I made this comment in the UC20 boot PR: https://github.com/snapcore/snapd/pull/7992#issuecomment-574609408 | 11:02 |
| mup | PR #7992: bootloader: add ExtractedRunKernelImageBootloader interface, implement in grub <UC20> <⛔ Blocked> <Created by anonymouse64> <https://github.com/snapcore/snapd/pull/7992> | 11:02 |
| * sdhd-sascha re | 11:03 | |
| pstolowski | sigh, 'git push merge master' should dwim | 11:03 |
| sdhd-sascha | zyga: what do you think, what is the best way to solve the issue with realpath ? Would be nice, if it would be possible to mount tmpfs over /dev/shm. So i didn't need the preload library with the realpath issue. | 11:05 |
| * Chipaca brb | 11:06 | |
| zyga | sdhd-sascha: is the problem that realpath() does not behave as it is documented to? | 11:07 |
| sdhd-sascha | zyga: with the preload library, realpath goes the call-path, like GLIBC <= 2.3. Without it like modern. The preload code, seems to choose a wrong entry point for realpath. | 11:13 |
| sdhd-sascha | I mean: with preload: it returns NULL instead of the path, if the seconde param is NULL. | 11:13 |
| zyga | sdhd-sascha: can this be fixed by changing the preload library? | 11:13 |
| zyga | sdhd-sascha: perhaps fixing it in snapcraft itself | 11:13 |
| sdhd-sascha | zyga: yes, i'm sure, it could be fixed by changing the preload library. But i don't know yet, how i could make this function call `dlsym (RTLD_NEXT, FUNC_NAME)` to choose the correct version. | 11:16 |
| sdhd-sascha | The preload-library isn't part of snapcraft, yet. https://github.com/sergiusens/snapcraft-preload | 11:17 |
| zyga | I see | 11:17 |
| zyga | sdhd-sascha: can you use dlvsym? | 11:17 |
| zyga | it allows you to pick the version you want | 11:18 |
| sdhd-sascha | zyga: well, i can think about it, to change all library calls... good, idea | 11:18 |
| sdhd-sascha | zyga: what glibc version would you prefer for the future ? | 11:20 |
| zyga | > | 11:20 |
| zyga | I have no idea what I am supposed to answer | 11:20 |
| sdhd-sascha | maybe, 2.19 ? | 11:21 |
| sdhd-sascha | zyga: thank you | 11:23 |
| zyga | no idea :) | 11:23 |
| zyga | can you explain why one over another? | 11:23 |
| zyga | perhaps you should dlvsym the newset that you knew about | 11:24 |
| zyga | and then fall back | 11:24 |
| zyga | note that it is not library version | 11:24 |
| zyga | but symbol version | 11:24 |
| zyga | check the symbol versions available for realpath | 11:24 |
| zyga | and probably pick the one that works the way you want | 11:24 |
| zyga | (one exact hit) | 11:24 |
| sdhd-sascha | zyga: no, i have to read the documentation. Many of the redirected calls are from linux itself, and i only need to redirect the stdlib, and so ... | 11:24 |
| * zyga doesn't understand how that is related | 11:25 | |
| zyga | or doesn't understand the original question | 11:25 |
| mup | PR snapd#8000 closed: release: 2.43.1 <Created by mvo5> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/8000> | 11:25 |
| sdhd-sascha | zyga: sorry. I think you understand me. My last sentence was about specific to serguissens code ;-) | 11:25 |
| sdhd-sascha | zyga: i ask about the version. Because i would create a PR, if i get it worked. I need at least > 2.3 | 11:26 |
| zyga | you can always reimplement realpath | 11:27 |
| zyga | if good version in glibc -> use it | 11:27 |
| zyga | otherwise -> here's one from scratch | 11:27 |
| sdhd-sascha | zyga: well. my problem is not realpath. It's /dev/shm. That's why i use the preload. But then the realpath didn't worked... | 11:28 |
| zyga | sdhd-sascha: what's with /dev/shm that is a problem for your app? | 11:28 |
| sdhd-sascha | zyga: yes. Apparmor says permission denied for /dev/shm/.... | 11:29 |
| zyga | right, because apps tend to drop random bits of shm there and we don't want them to use it as unmediated IPC | 11:29 |
| sdhd-sascha | Yes, and here, i first patch sway to use "mktemp...". Then the "application menu" wants to write /dev/shm too... | 11:30 |
| zyga | interplay of apps and /dev/shm is ... complex | 11:31 |
| sdhd-sascha | I mean, i patched "wlroots", which i base library for wayland desktop's. | 11:31 |
| sdhd-sascha | is | 11:31 |
| sdhd-sascha | It would be nice, if a snap could mount tmpfs over /dev/shm. But is then with xdg-proxy, i some application want to talk | 11:33 |
| sdhd-sascha | What is with xdg-proxy... I mean "inter-snap" communication of snaps over "/dev/shm" | 11:34 |
| sdhd-sascha | zyga: you already helped me. I think i try to fix the preload-library ?! | 11:35 |
| zyga | sdhd-sascha: I think that usage of /dev/shm must be changed and reorganized | 11:37 |
| zyga | sdhd-sascha: currently it's pretty much like exchanging files via /tmp | 11:37 |
| sdhd-sascha | Oh, well... What if i use dlvsym with version... And then there are some more preloads used | 11:37 |
| zyga | sdhd-sascha: we don't have private per-snap /dev/shm because that would break everything | 11:37 |
| zyga | sdhd-sascha: but I think something along those lines would help a _lot_ | 11:38 |
| sdhd-sascha | Maybe, for my case is to recompile glibc, without the old `realpath`. Or, like you says. Write another `realpath`. Or i patch sway, to use the old calling convention | 11:39 |
| zyga | what I was describing was a long term plan | 11:40 |
| zyga | for short term we need hacks and hacks and tweaks to apparmor policy | 11:40 |
| zyga | sadly :/ | 11:40 |
| sdhd-sascha | I think, something like chroot and overlayfs would change the snap world. | 11:41 |
| zyga | you can use chroot | 11:41 |
| zyga | but I don't think that helps with the complexity of modern IPC | 11:41 |
| Chipaca | sdhd-sascha: do you have a minimal snapcraft with a .c that tries to use realpath and fails? | 11:41 |
| zyga | wayland + pulseaudio == loads of /dev/shm usage | 11:41 |
| sdhd-sascha | This code, fails on local machine, too. If i use the LD_PRELOAD=./libsnapcraft-preload.so | 11:43 |
| sdhd-sascha | https://www.irccloud.com/pastebin/GubDCCNh/ | 11:43 |
| sdhd-sascha | on, eoan | 11:43 |
| mborzecki | cachio: hi, can you trigger an update of debian-sid base image? | 11:44 |
| mborzecki | cachio: apt in that image needs to be updated, otherwise this problem can happen occassionally https://paste.ubuntu.com/p/WVGk8pW7rF/ | 11:45 |
| mvo | I guess "apt full-upgrade -y" is not a bad idea | 11:46 |
| sdhd-sascha | zyga: the code above, goes in gdb the way to "glibc-2.30"/stdlib/canonicalize.c" : | 11:47 |
| cachio | mborzecki, hello, sure, let me take a look | 11:48 |
| sdhd-sascha | https://www.irccloud.com/pastebin/kL1fM19L/ | 11:48 |
| sdhd-sascha | And there, it's returns NULL as path | 11:48 |
| sdhd-sascha | If i use, dlvsym, then no other preload can overwrite realpath. | 11:50 |
| sdhd-sascha | Or, i'm wrong ? not sure | 11:50 |
| zyga | sdhd-sascha: I dont think that works this way | 11:51 |
| zyga | sdhd-sascha: but I cannot focus on this right now, sorry | 11:51 |
| zyga | sdhd-sascha: dlvsym is really just "in the symbols I've loaded, find the one with the version marker as well" | 11:52 |
| zyga | it's not "load just that version" | 11:52 |
| zyga | you really only open one glibc.so | 11:52 |
| zyga | inside you look for a symbol with matching name *and* version | 11:52 |
| zyga | that's the trick | 11:52 |
| zyga | IMO | 11:52 |
| cachio | mborzecki, https://travis-ci.org/snapcore/spread-cron/builds/636199611 | 11:52 |
| cachio | mborzecki, it was updated on Monday | 11:53 |
| cachio | the image we use for our tests | 11:53 |
| sdhd-sascha | zyga: thank you. I was wrong. It's still possible to use other preloads, because of the loading order of the libs ;-) | 11:54 |
| mup | PR snapd#7986 closed: store, o/snapstate: send default-tracks header, use RedirectChannel <Created by chipaca> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/7986> | 11:54 |
| mup | PR snapd#8004 opened: cmd/snap: print full channel in 'snap list' <Created by chipaca> <https://github.com/snapcore/snapd/pull/8004> | 11:57 |
| Chipaca | pedronis: do we want snap info to explicitly point out a default track? | 12:02 |
| Chipaca | as in, default-track: foo if foo != "" | 12:02 |
| mborzecki | cachio: so maybe we need to do a full-upgrade like mvo indicated, apt version in sid is 1.8.4 atm, but our image has 1.8.0~alpha3 for some reason | 12:04 |
| cachio | mborzecki, ok | 12:04 |
| cachio | mborzecki, we are not doing the full upgrade for sid because of this | 12:05 |
| cachio | # We are not making dist-upgrade because it is creating a dependencies issue | 12:05 |
| cachio | # which is making fail the preparation of the snapd test suite | 12:05 |
| cachio | I'll manually trigger and then we can see how it works | 12:05 |
| cachio | mborzecki, running | 12:07 |
| cachio | in case the image breaks we can revert it | 12:07 |
| zyga | brb. | 12:17 |
| mvo | cachio: aha, this is sid? yeah, then feel free to just uprade apt | 12:20 |
| cachio | mvo, I updated the file and am upgrading sid, it will be ready for testing in few minutes | 12:21 |
| mvo | cachio: nice, thank you | 12:22 |
| mborzecki | cachio: thanks! | 12:27 |
| pedronis | Chipaca: there was this proposal to put a default note on the most stable risk of it | 12:38 |
| pedronis | Chipaca: if we add default-track: it should probably be only in --verbose | 12:39 |
| Chipaca | pedronis: foo/candidate ♪ :) | 12:44 |
| pedronis | Chipaca: this re-landed only recently right: https://forum.snapcraft.io/t/behavior-change-risk-only-channel-specifications-will-not-switch-track/11769 , it's not in 2.43 ? cc mvo | 12:46 |
| pedronis | context: https://forum.snapcraft.io/t/the-snapd-roadmap/1973 | 12:47 |
| mvo | ijohnson: sure thing, we are in the standup ho now | 14:59 |
| Chipaca | failures in debian-sid are expected right now, yes? | 15:01 |
| sergiusens | pstolowski: hey there, mind taking a look at https://github.com/snapcore/snapcraft/pull/2876 | 15:14 |
| mup | PR snapcraft#2876: meta: handle plug & slot string objects <Created by cjp256> <https://github.com/snapcore/snapcraft/pull/2876> | 15:14 |
| pstolowski | sergiusens: sure, with a caveat that my python foo got a bit rusty ;) | 15:17 |
| pstolowski | but it's better than my vala foo ;) | 15:18 |
| * zyga still fights with a bug | 15:39 | |
| zyga | I should break for coffee and think instead of just hammering the keyboard | 15:39 |
| sergiusens | pstolowski: insights appreciated more around the semantics around slots/plugs than the code :-) | 16:02 |
| ijohnson | mvo: should we add `system-backup` (and potentially any other new interfaces in 2.43) to the snapd release roadmap as new features for 2.43? | 16:24 |
| mvo | ijohnson: yes! feel free to edit | 16:30 |
| ijohnson | mvo: cool :-) | 16:30 |
| === pedronis_ is now known as pedronis | ||
| pedronis | Chipaca: does our snap/partial caching do nothing in this case? https://forum.snapcraft.io/t/failure-to-fetch-assertions-stops-installation-immediately/15059 | 16:38 |
| Chipaca | pedronis: no | 16:40 |
| Chipaca | pedronis: not if the change was aborted | 16:40 |
| pstolowski | sergiusens: reviewed | 16:46 |
| sergiusens | thanks pstolowski | 16:46 |
| pstolowski | yw | 16:47 |
| mup | PR snapcraft#2876 closed: meta: handle plug & slot string objects <Created by cjp256> <Merged by sergiusens> <https://github.com/snapcore/snapcraft/pull/2876> | 16:48 |
| * cachio lunch | 17:01 | |
| mup | PR snapd#8005 opened: api: don't return connections referring to non-existing plugs/slots <Created by stolowski> <https://github.com/snapcore/snapd/pull/8005> | 17:24 |
| mup | PR snapcraft#2878 opened: extensions: fix merge_lists to handle non-string list cases <Created by cjp256> <https://github.com/snapcore/snapcraft/pull/2878> | 17:39 |
| mup | PR pc-amd64-gadget#33 opened: grub.cfg-boot: use $prefix/kernel instead of $ubuntu-boot/kernel <Created by anonymouse64> <https://github.com/snapcore/pc-amd64-gadget/pull/33> | 17:39 |
| mup | PR snapcraft#2879 opened: [3.9-backport] meta: ensure Snap's `assumes` is initialized as a set <Created by cjp256> <https://github.com/snapcore/snapcraft/pull/2879> | 18:21 |
| mup | PR snapd#8006 opened: tests: skip interfaces-network-manager on arm devices <Created by sergiocazzolato> <https://github.com/snapcore/snapd/pull/8006> | 19:00 |
| pedronis | ijohnson: do we still need Recovery: true in the new tests in 7992 | 19:01 |
| mup | PR snapd#8007 opened: tests: remove execution of ubuntu 19.04 from google backend <Simple 😃> <Created by sergiocazzolato> <https://github.com/snapcore/snapd/pull/8007> | 19:11 |
| === zylop is now known as ExplicitedE_Xplo | ||
| === ExplicitedE_Xplo is now known as Zylop | ||
| === Zylop is now known as zylop` | ||
| mup | Bug #1745772 changed: Can't run snap in lxd container <Snappy:Fix Released> <https://launchpad.net/bugs/1745772> | 21:41 |
| mup | Bug #1664427 changed: thefuck snap gets an apparmor denial even in classic confinement <classic> <isv> <Snappy:Confirmed> <https://launchpad.net/bugs/1664427> | 21:44 |
| mup | Bug #1675413 changed: snap disconnect doesn't support tab completion <Snappy:Fix Released> <https://launchpad.net/bugs/1675413> | 21:44 |
| mup | PR snapd#8008 opened: render: add the render package and basic widgets <Created by zyga> <https://github.com/snapcore/snapd/pull/8008> | 23:02 |
| mup | PR snapcraft#2880 opened: [RFC] schema: introduce configuration for package management repositories <Created by cjp256> <https://github.com/snapcore/snapcraft/pull/2880> | 23:22 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!