[01:57] * mwhudson pokes at glibc's build system, fails === ben_r_ is now known as ben_r === pieq_ is now known as pieq [09:02] rbasak: any idea why "git ubuntu tag --upload" would complain about "ERROR:Working tree must be clean to continue." while "git status --ignored" is empty? [09:03] ah, could it be that it has an empty dir issue? [09:03] I just checked out my current branch and got the "WARNING: empty directories exist but are not tracked by git:" ... [09:04] rbasak: by that the upload won't be matched anyway, would that also trigger the "must be clean" issue [09:21] I'm not sure, sorry. That behaviour always seemed a little buggy to me. [09:23] what confuses me is that it depends on the type of tag [09:24] --split almost always works whiel --logical almost never works (for me) [10:40] doko: would you mind reviewing bug 1864766 please? It looks like the right thing to do to me, but I'd appreciate your eyes. [10:40] bug 1864766 in python-pip (Ubuntu Xenial) "[SRU] pip in xenial is installing packages incompatible with Python 2.7 (and those are becoming common)" [Medium,In progress] https://launchpad.net/bugs/1864766 [11:07] Morning/afternoon [11:09] After update from eoan to focal, ina XPS 13, X11 simply stop to proper work: [11:09] Here's what is happening https://photos.app.goo.gl/MPv8Ef7BnUYA7VXb9 [11:09] Right now i'm using wayland [11:09] Is there any report open against intel graphic drivers ? === bpsecret- is now known as bpsecret [11:18] hmm, the snap version of gnome-software on latest Ubuntu Focal does not uninstall unused dependencies when uninstalling a deb application? [11:18] or is there something wrong on my system? [11:20] also, I was able to uninstall the Snap Store snap itself from the Snap Store app... this should not be possible at all [11:22] I am not sure if I should report these as issues on Launchpad since they happen only in the snap version of gnome-software [11:24] oh and another small thing: the "Open in Software" button from Applications tab in GNOME Settings does nothing when there is only a Snap version of gnome-software installed === Wryhder is now known as Lucas_Gray [12:03] AsciiWolf: https://snapcraft.io/snap-store says to use https://bugs.launchpad.net/snap-store/ [12:08] rbasak, thanks! [12:08] hi, I am investigating the gem2deb/1.0.5 autopkgtest failure on armhf and it is complaining (reprotest actually) about the kernel version: https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-focal/focal/armhf/g/gem2deb/20200311_020802_c6e4e@/log.gz [12:09] if you check the log you will find: autopkgtest [01:38:52]: testbed running kernel: Linux 4.15.0-72-generic #81-Ubuntu SMP Tue Nov 26 12:21:09 UTC 2019 [12:09] is linux 4.15.0 expected in an ubuntu focal testbed? [13:22] vorlon: hi, for when you are online, when you have a moment, this is a seed change for a geoip library. I added a review slot for you at the request of the MIR team, because you were the one who added the geoip1 version to this seed way back: https://code.launchpad.net/~ahasenack/ubuntu-seeds/+git/ubuntu/+merge/380547 [13:26] teward: hi, I'm in the process of trying to bring libmaxminddb (for GeoIP2 access) from universe to main, and demote geoip1. There are only two rdeps in main, and nginx is one of them. I believe I just need to reshuffle the meta packages to require the geoip2 subpackage instead of the geoip1, and I'll put an MP up shortly and add a review slot for you [13:52] teward: when you are around, for the moment the diff is in https://bugs.launchpad.net/ubuntu/+source/libmaxminddb/+bug/1861101/comments/8 as I can't push to a new git repo in lp (it's being worked on) [13:52] Launchpad bug 1861101 in nginx (Ubuntu) "[MIR]: dependency of bind9" [Undecided,In progress] [13:57] rafaeldtinoco: will your fix for bug 1866458 definitely also work on the release pocket Bionic kernel? Because we'll build with NLA_F_NESTED defined to non-zero, right? [13:57] bug 1866458 in drbd-utils (Ubuntu Bionic) "drbd not working after kernel upgrade 5.0.x -> 5.3.x" [Medium,In progress] https://launchpad.net/bugs/1866458 [13:58] yep [13:58] without that flag set command rets error [14:02] rafaeldtinoco: I don't follow [14:02] I thought this worked without the fix on the original kernel? [14:03] rbasak: new kernels containing a specific commit need that fix [14:03] so if user is using HWE and upgrades to new kernel.. might have to have it [14:04] ah I see what you mean [14:04] sorry rbasak I was doing 3 simult things [14:04] you are afraid of the flag [14:04] being added to a kernel that does not need it [14:04] +#ifndef NLA_F_NESTED [14:04] +#define NLA_F_NESTED 0 [14:04] +#endif [14:05] that part will make it ok [14:06] or maybe I didnt understand what u meant [14:06] rafaeldtinoco: sort of [14:06] And yes, that ifndef would nearly make me happy [14:06] However that's a build time thing [14:07] In our case, presumably NLA_F_NESTED is always available at build time and is non-zero [14:08] I see what u mean [14:08] i was checking that [14:08] So how does something compiled with that work with an old kernel being given a flag value it doesn't recognize? [14:08] linux/netlink.h:#define NLA_F_NESTED (1 << 15) [14:08] * rafaeldtinoco checks [14:12] rbasak: https://bugs.launchpad.net/ubuntu/+source/drbd-utils/+bug/1866458/comments/11 [14:12] Launchpad bug 1866458 in drbd-utils (Ubuntu Bionic) "drbd not working after kernel upgrade 5.0.x -> 5.3.x" [Medium,In progress] [14:13] rafaeldtinoco: thank you for checking! [14:13] of course. im sorry for not having checked it before [14:14] ill pay more attention to compile-time only variables being added [14:14] It's fine. That's the point of reviews :) [14:14] yep [14:14] =) === Wryhder is now known as Lucas_Gray [15:36] vorlon, hey, could you get wpebackend-fdo to build on i386? it's a build-depends of webkit2gtk now [15:39] we need mozjs68 for gjs now as well [15:42] https://code.launchpad.net/~laney/ubuntu-archive-tools/mozjs68/+merge/380562 [15:49] seb128: wpebackend-fdo added to the whitelist. so webkit2gtk got uploaded while there was a version entangled in the icu transition, and there were autopkgtest failures on webkit2gtk's revdeps in particular that were blocking. are we going to get this through? [15:50] vorlon, I noticed those but I didn't want to investage on an rc version when a new stable is out which might fix it, I will look at those failure if there are still there with 2.28.0 [15:51] ok, as long as there's follow-through, thanks [15:51] np! [15:59] ahasenack: NACK unless your rejiggered the NGINX code. the inbuilt GeoIP doesnt work with geoip2 [16:00] we had to pull in a third party plugin for it so unless libmaxmind is going to Main I have to disable some components on NGINX builds [16:01] I have no issues doing that but that removes the geoip functions from nginx-core [16:01] had this discussion with upstream when GeoIP Legacy went the way of the dinosaur and we pulled in geoip2 as another third party module [16:02] ah wait I misread. bringing libmaxmind to main. [16:02] error:deadbrain [16:02] ahasenack: sec team needs to review geoip2 then [16:03] because that's a third party module [16:03] (the geoip2 in the package for nginx) - if Security OKs that third party module going to main then I can apply it [16:03] teward: you mean review nginx's geoip2 core, not libmaxminddb [16:03] ahasenack: ehhhhhhhhhhhh [16:03] no [16:03] is that a separate source tarball? Or a contrib/? [16:04] hang on a moment IRC from phone is inefficient [16:04] let me get to my PC [16:04] teward: I'm on a call too :) [16:04] teward: we can talk in 1h or later [16:04] lol i"m not heh [16:04] FWIW I switched Launchpad to geoip2 in production a while back [16:04] just not out of bed :) [16:05] (because geoip legacy is (a) legacy and (b) doesn't support ipv6) [16:07] cjwatson: right, that's why i pushed to get geoip2 added to the package as a third party dep [16:07] but here's the problem [16:07] https://bugs.launchpad.net/ubuntu/+source/nginx/+bug/1825895 is the inclusion of geoip2 module [16:07] Launchpad bug 1825895 in nginx (Ubuntu) "Please include GeoIP2 module." [Wishlist,Fix released] [16:07] which is not from nginx.org upstream [16:08] it's in debian/modules/ [16:08] which is where we load 3rd party modules (Debian's to blame here) [16:08] now, we had a strict thing with nginx-core to not have any third party modules back in the nginx MIR [16:08] teward: I have to leave for about 1h, will be back later to read history [16:08] geoip is from upstream, geoip2 is *not* from upstream nginx and is third party [16:09] so if we're going to add that as a dep in nginx-core we need the SEcurity team to vet that module's code more thoroughly [16:09] (cc sarnold because I'm about to give you more work_) [16:09] now, if the Security team ACK's the geoip2 module that is third party for Main, then I can support this and push the diffs in [16:09] I presume this is only an issue due to the desire to only have one or the other of the core geoip{1,2} libraries in main [16:10] since I don't think nginx uses the python geoip bindings either way [16:10] anyway, not my problem, I just wanted to mention the Launchpad change [16:13] cjwatson: the only reason it's an issue is because ahasenack wants to change nginx-core to pull in libnginx-mod-http-geoip2 instead of libnginx-mod-http-geoip [16:13] Presumably ultimately for the reason I gave [16:13] ahasenack: this said, the two can work at the same time, they're designed to be coinstallable and work with slightly different commands [16:13] yep [16:13] cjwatson: yep. [16:15] regardless, because of the nature of -geoip2 being a third party module that was NOT ACK'd in the original NGINX MIR (and was not given an ACK for main because libmaxminddb is not in Main) it needs a Security Team review of that code before it can be ACK'd and included in Main (the -geoip2 module). [16:15] s/that code/the third party module's code/ [16:17] (it also required libmaxminddb0 to be installed at runtime so... hence why it couldn't be in Main directly) [16:18] ahasenack: also, I would suggest that we keep -geoip in as well as -geoip2 because -geoip2 isn't AIUI reverse compatible with GeoLite or GeoIP1 which some older deployments may still be using. [16:18] (because they target different geoip datasets and there's a LOT of legacy apps I can think of in the Universe that use the legacy geoip data) [16:21] ah i just read the bug data. ACK for dropping -geoip as a dep in core IF security team ACKs -geoip2 for Main [16:21] cpaelzer: ^ pulling you in since you pointed ahasenack my way [16:22] cpaelzer: needs security team reivew of -geoip2 for Main inclusion before it can move anywhere [16:22] (because -geoip2 is NOT from nginx upstream, it's a third party contrib.) [16:22] I feel like you've said this a lot and maybe it would be better to just write it down once in a bug or MP or whatever's appropriate? [16:22] already made a note in the bug [16:23] *yawn* now where the heck is my coffee... [16:33] Laney: well, mozjs68 ftbfs on i386 :P [16:33] Laney: (and on amd64 apparently) [16:33] WHAT [16:33] all I did was a no-change rebuild to pick up i386 [16:34] so I dunno [16:34] dunno about amd64, no log? [16:35] i386... yeah... ok that looks plausibly real [17:19] teward: I commented in the bug also. What if we demoted the geoip(1) nginx module? [17:21] ahasenack: doable. would require a few tweaks to the build rules too I believe because it references the dynamic modules there. give me a bit, doing some net maintenance at home atm so stuck on my phone's crap signal. Release Notes will need a blurb then though [17:21] sure [17:22] so upgraders know the functionality changed [17:22] I added some reasons to my comment [17:22] which we could use to justify this [17:22] *disappears to finish the net changes to bring wifi back up* [17:22] good luck! [17:39] ahasenack: given the MIR backlog currently at the Sec Team's level (per sarnold) I'd say just drop geoip from core. I'll have to tweak the build rules give me a bit to poke some things and I'll have a debdiff [17:39] thanks teward. This also doesn't prevent the review of geoip2 from happening [17:39] correct. [17:40] but it removes NGINX from the equation [17:40] for now [17:40] agreed [17:40] we can MIR geoip2 in nginx later [17:40] if you want I can file that bug and start the analysis, like I did for bind9 (see the long description) [17:41] yep, feel free, but let me get this diff first and let you once-over it to see if I missed stuff [17:42] right, let's see how it works [17:42] and if it works [17:43] woo, thanks teward, ahasenack :) [17:43] teamwork++ [17:49] ahasenack: https://paste.ubuntu.com/p/4q3Bv9Sy53/ <-- the debdiff I've got that would make these changes. [17:49] there's an extra change in there for consistency sake on indentation from the 'dynamic module' section defining what is/isn't a dynamic module but the core_configure_flags in debian/rules is the core change that needed to be made [17:49] otherwise it'd explode at compile :P [17:50] * ahasenack doesn't like explosions outside of cinema [17:51] teward: I'll give that a whirl in a ppa together with the rest, and MP afterwards [17:51] unless you are doing that right now [17:51] nah i need foods. feel free to do the PPA testing, if it works i can upload this direct :P [17:51] ok [17:51] thanks a lot [18:47] wow LAG. my debdiff post JUST went throug bleh [18:47] *shoots computer* [18:55] ahasenack: i just uploaded the debdiff as well [18:55] to the bug in case paste.u.c drops [18:59] teward: my only change is a whitespace at the and of the changelog message, and adding the bug number [18:59] cool cook [18:59] + * Drop GeoIP from nginx-core due to demotion of libgeoip (LP: #1861101): [18:59] Launchpad bug 1861101 in nginx (Ubuntu) "[MIR]: dependency of bind9" [Undecided,In progress] https://launchpad.net/bugs/1861101 [18:59] cool cool* [18:59] it's building in a ppa [19:00] awesome let me know how it goes [19:00] https://launchpad.net/~ahasenack/+archive/ubuntu/bind9-geoip [19:13] ahasenack: I'll run tests after it builds and publishes [19:13] it's published, I'm about to install it [19:13] tw^ [19:13] teward: ^ [19:13] ok. i need to check some config bits tho on it too [19:13] make sure its disabled too - because default it's off :P [19:13] in a few once my Focal container starts [19:17] teward: note I enabled proposed in the ppa, to mimic what will happen once it's uploaded to focal [19:17] awesome [19:17] there is a new libcrypt1 dep [19:17] you might have to enable proposed in your container as well [19:17] (I already do heh) [19:17] yay [19:18] nginx-core : Depends: libcrypt1 (>= 1:4.1.0) but it is not installable [19:18] I didn't :P [19:23] I see it there, it has the so, linked to the expected libraries [19:23] teward: modules-available/ is empty after installs? [19:23] in /etc, I mean [19:23] yes [19:25] ahasenack: mine's enabled too but it won't install until i specify `-t focal-proposed` to apt :P [19:25] installing now [19:25] safer [19:26] (I pinned it at 400 per https://wiki.ubuntu.com/Testing/EnableProposed) [19:26] I always forget about -t [19:27] ahasenack: modules-enabled is where the system installed ones are [19:28] geoip isn't in there which is a good sign [19:28] but let me do some tests [19:29] nginx: [emerg] unknown directive "geoip_country" in /etc/nginx/conf.d/geoip.conf:1 <-- this is *good* [19:29] means geoip isn't available which is what our proposal does to nginx-core - removes 'geoip' from the available command set [19:29] so from my perspective LGTM with those changes in, as geoip isn't available [19:30] that said, we'll start getting [emerg] level log warnings for people still using legacy GeoIP in nginx-core and upgrading [19:30] but that's something we'll have to justify [19:33] is this something that prevents nginx from starting after an upgrade? [19:33] I guess no, because we are not removing the geoip package [19:33] it will just be upgraded as usual, but from universe [19:33] this was a fresh install case? [19:35] yes [19:35] but the upgrade case will cause errors as well [19:35] how so? [19:35] if someone doesn't have universe enabled, you mean? [19:35] well if the upgrade process does an autoremove it'll remove the geoip module [19:36] and its modules-enabled reference [19:36] and restart nginx and get an 'unknown directive' error [19:36] this'll require a release notes notice about it [19:36] will it fail to restart? [19:36] yes [19:36] I think it prompts for autoremove [19:36] but still, if there are many packages in that list, it might be missed [19:36] yep [19:37] if it autoremoves libnginx-mod-http-geoip it'll error with the error I indicated [19:37] replacing geoip with geoip2 would have caused the same thing, no? [19:37] correct [19:37] because it's a separate set of config arguments [19:38] this is ExpectedIssues with any major change like this [19:38] I mean, even the original plan would have caused this [19:38] correct [19:38] ok, document and proceed. I'll add a release notes task to the bug [19:38] if I remember what the release notes project is [19:38] heh [19:40] ahasenack: i'll tweak my debdiff, do you want me to upload direct to proposed then? [19:40] teward: that /etc/nginx/conf.d/geoip.conf you had, was that you who created it, or the package? [19:41] i created it [19:41] ok [19:41] no package makes anything in conf.d [19:41] so even purging would leave it there, and the error [19:41] but it beats the heck out of editing nginx.conf directly (it has an `include /etc/nginx/conf.d/*.conf`) [19:41] correct, which is another example of how this upgrade can break things [19:41] yeah, isn't that one of the reasons to have .d? To have packages place stuff there? [19:41] ahasenack: indeed, but nginx doesn't enable geoip by default [19:41] upstream indicates as much [19:42] but if you install it, it loads it by default [19:42] http://nginx.org/en/docs/http/ngx_http_geoip_module.html [19:42] just not uses it? [19:42] correct [19:42] but loads vs. uses is the point here [19:42] ok, several layers [19:42] the module is default off unless you configure things [19:42] once you add the configuration (a-la that link's example at the top of its docs) it tries to use the module [19:42] by 'default off' i mean enabled anda vailable but not used. [19:42] s/enabled/loaded/ [19:43] just like a2enmod {INSERTMODULE} that doesn't necessarily *activate* any configurations for some modules just makes it available for conf to use. [19:43] (in some edge cases) [19:43] ok [19:44] I added a comment about release notes [19:44] wrt uploading now, I don't know, cpaelzer isn't aware of our new plan [19:44] maybe wait until tomorrow? [19:44] ack [19:44] i'll prep the changes on my side though [19:44] just in case ;) [19:44] ok, thanks [19:44] I can't create an MP, launchpad is timing out when I git push [19:45] known issue, I raised a support ticket [19:45] right. [19:45] i don't push to the Git with nginx, though, I usually upload direct. [19:45] (old school package - I haven't gotten a git workflow for it worked yet) [19:45] s/package/packaging/ [19:45] it's fine [19:45] unlike my GBP workflow in Debian for vmfs6-tools and xca :P [19:46] teward: what tz are you in? [19:46] it's 16h46 for me now [19:46] (19h46 UTC) [19:47] Eastern US (15:46 local, 19:46 UTC) [19:47] +- 1m30s for typing resulting in timeoffset [19:47] oh, look, same utc [19:47] :D [19:47] close enough [19:47] yep xD [19:48] LocutusOfBorg, I mistakenly sync'ed a new version of src:puma on top of your delta and now I am trying to introduce the changes again to fix a regression. However, some questions arose during the review process, could you check this bug and try to give us some insight? https://bugs.launchpad.net/ubuntu/+source/puma/+bug/1866881 [19:48] Launchpad bug 1866881 in puma (Ubuntu) "Autopkgtest regression triggered by Ruby 2.7" [Undecided,New] [19:49] teward: thanks for your assistance on this today [19:49] ahasenack: you're welcome. do me a solid though and prep the MIR for -geoip2 module as well? [19:49] we'll throw that stuff onto Security's backlog ;) [19:50] if not in time for release here, for G-series [19:50] already created the card: https://trello.com/c/ySFJNH0x [19:50] nice [19:50] you can subscribe to it, it will have the mir bug shortly [19:51] yep watching the card now :) [19:52] and i'm subbed to all nginx bugs so when that MIR is filed I'll get a ping :) [19:53] sorry I made things complex but :) [19:53] best we're all in the loop :) [20:00] you did not make it comple [20:00] complex [20:00] I missed the libmaxminddb requirement in my bind9 9.16 update, I thought I could still use the old geoip1 [20:08] vorlon: I ran across this MP the other day - do you have an opinion on the new sizes? https://code.launchpad.net/~xnox/ubuntu-release-upgrader/eoan-kernel-sizes/+merge/374125 [20:17] kanashiro, will have a look [20:21] bdmurray: seems reasonable to me, particularly as these are fallbacks [20:39] kanashiro, I did sponsor the patch thanks [20:43] LocutusOfBorg, thank you [21:01] kanashiro, I added also the previous changelog entries [21:01] so we don't loose the history [22:06] vorlon: bah ubuntu-dev-tools in focal is not compatible with python2 any more [22:06] heh [22:07] sil2100: pleeease [22:07] i'll poke a bit to see how non-trivially [22:55] bluuurgh python2 sucks