[00:09] rbasak: cpaelzer: i'm killing the importer cron job until i can debug this better -- i'm seeing different results from the import-cron script and the git-ubuntu invocation, which shouldn't happen [00:23] nacc: ack [01:41] cpaelzer: i pushed --reimport to the snap, if you refresh it should be available and i think will work for you [02:32] andol: yes. effective processing of the step isnt fruitful [07:12] Good morning [07:13] morning [07:19] Hey TafThorne [07:33] hello to you too lordievader === lifeless_ is now known as lifeless [07:57] cpaelzer: morning - shall I send you a git review with my workaround patch for the double-locking issue in libvirt? I'd like to get this resolved AM today if possible [08:14] jamespage: hmm, not sure yet [08:14] jamespage: unfortunately upstream didn't respond at all yet [08:14] jamespage: but we can "hotfix" it now and reroll to the upstream solution later [08:15] cpaelzer: +1 we need todo that [08:15] jamespage: I can take the diff from the LP bug [08:15] jamespage: were there any updates to is since you attached? [08:15] cpaelzer: nope that's the one we tested with last night [08:15] result? [08:15] good? [08:16] cpaelzer: yes - commented to that effect on the bug just now [08:16] perfect I'll take over [08:17] cpaelzer: thanks! [08:17] review, build, test and ping you then [08:17] AM for the uplaod maybe, but to migrate I'm not so sure [08:26] jamespage: so you ended up not moving the dmn->autoShutdownCallingInhibit = true; that I asked about [08:26] jamespage: and only do the "hotfix as a partial revert [08:27] cpaelzer: yeah basically - by switching back to the older behaviour, that call is always good [08:27] which might return what the former patch fixed but seems better than what we have for now [08:27] ack [08:27] as hotfix ok [08:28] cpaelzer: agreed - its needs a more complete fix if upstream want to maintain that fast-path behaviour [08:47] cpaelzer: upstream fix also looks reasonable btw [08:57] jamespage: already building with it [08:57] jamespage: https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/2942 [08:58] jamespage: for a fast minor sniff before pushing [08:58] jamespage: I'll give you a ping when it is ready to test for you as well [08:58] jamespage: or would you need the xenial backport to verify your case? [09:26] cpaelzer: ideally but I can deal with that [09:55] jamespage: https://launchpad.net/ubuntu/+source/libvirt/3.6.0-1ubuntu4 [09:55] jamespage: strictly speaking it is AM :-) [09:56] jamespage: yet we need to guide it through migration [09:56] jamespage: this is already the upstream fix [09:59] you even have 1 extra hour of AM left for you [10:55] cpaelzer: have it built - but had some image issues in our cloud so only just re-deploying... [10:55] cpaelzer: apologies [11:02] jamespage: all fine, let me know how the retests on this fix behave [11:05] cpaelzer: tbh unless you see a regression I'd go with that patch for upload to arftul [11:05] cpaelzer: it looks 100% reasonable and correct to me [11:10] smoser: I owe you a simplestreams change for a niggle I need to scratch [11:11] smoser: I'd like to split 'keep' from 'maxnum' in resolve_items (optionally) [11:11] as its desirable to start seeding a cloud with a smaller number of image syncs, but retain synced images over time [11:12] it looks like keep may have done that at some point in the past, but its a boolean toggle now and will only keep up to maxnum [11:39] anyone know of a small cmd line tool that can create sha512 crypt passwords? Dont really want to install dovecot for dovecotpw. [11:47] albech: sha512? [11:48] As in that is what it comes up with on my command lines. I think it is a thing out of openssl dgst [11:52] cpaelzer: looping a reboot --hard right now [11:52] looking good so far [11:53] Oh I was meant to try out something for fsck now I modified /etc/fstab. Time for me to reboot too. [12:08] cpaelzer: +1 on that fix - lots-of-reboots later still all good [12:11] fsck finished suspiciously fast (few seconds on a 1.8 TB spinning drive) but it passed. [12:19] cpaelzer: I've pushed from the bileto ppa to artful - thanks for preparing that update [12:21] nm I see you already did that... [12:21] doh [12:33] jamespage, hm... yeah. keep is supposed to be separate from maxnum. [12:34] zlthough originally i think it was only to be less. keep being less than maxnum. [12:34] have to think again. [14:36] how would i nopte down that this remains unfixed for xenial while it is fixed in yakkety? https://bugs.launchpad.net/ubuntu/+source/mdadm/+bug/1568097 [14:36] Launchpad bug 1568097 in mdadm (Ubuntu) "Error messages about mdadm --incremental failing" [High,Fix released] [14:38] tomreyn: you can nominate it for xenial [14:38] other bug reports have these clock (?) icons here listing xenial, yakkety and zety as separate targets, e.g. https://bugs.launchpad.net/ubuntu/+source/mdadm/+bug/1608495 [14:38] Launchpad bug 1608495 in mdadm (Ubuntu Zesty) "IMSM fakeraid handled by mdadm: unclean mounted volumes on shutdown/reboot" [Critical,Confirmed] [14:38] but i can't seem to find out how to add these nominations on https://bugs.launchpad.net/ubuntu/+source/mdadm/+bug/1568097 [14:38] Launchpad bug 1568097 in mdadm (Ubuntu) "Error messages about mdadm --incremental failing" [High,Fix released] [14:38] tomreyn: there is a "nominate for series" link [14:38] "Also affects project (?) Also affects distribution/package Nominate for series" under the task list [14:39] or maybe you don't have it? [14:39] "[X] Also affects project" and "[X] Also affects distribution/package" is what i have [14:39] using a standard ubuntu account without any special permissions [14:41] ahasenack: since you have this option, could you do it? [14:42] (or more precisely, could you please evaluate whether it needs to be done) [14:45] cpaelzer: you didn't try again with --reimport, did you? (re: src:virt-manager) [15:11] cpaelzer: if you're able to join standup early, we can talk about that [15:37] rbasak: so i'm thinking we need to adjust something -- --no-fetch no longer implies --no-push if --no-fetch --directory is how you reuse adirectory? === sammyg is now known as FBI === FBI is now known as Guest15393 === Guest15393 is now known as sammyg [16:30] cpaelzer: virt-manager reimported and you can see that your upload tags were preserved: https://git.launchpad.net/~usd-import-team/ubuntu/+source/virt-manager/log/ [16:30] cpaelzer: will check on the MPs [16:30] cpaelzer: interesting, they got deleted -- can you resubmit? i'll adjust the code [16:46] beisner: hi, can you promote nova - 2:14.0.7-0ubuntu2~cloud1 to newton-proposed? [17:07] ahasenack: can you try your samba build with the latest snap? [17:07] I sure can [17:08] 211? [17:08] ahasenack: yeah, i think so (211/212 depending on arch) [17:09] nacc: is build-source aborting now when the tree is dirty? [17:09] ahasenack: yes [17:09] hm [17:09] ahasenack: because pristine-tar doesn't like that [17:09] ahasenack: to be clear, it was before too [17:09] I change the version in d/changelog to add a ~ppa1 just before a build, so I can upload to a ppa [17:09] ahasenack: (before == yesterday) [17:09] no, before it complained, but continued [17:09] ok, earlier then [17:09] ahasenack: yeah, earlier than yesterday :) [17:10] nacc: ok, didn't work [17:10] 09/01/2017 14:10:14 - ERROR:stderr: dpkg-buildpackage: error: fakeroot not found, either install the fakeroot [17:10] nacc: full output: http://pastebin.ubuntu.com/25446047/ [17:11] ahasenack: is that branch pushed somewhere? [17:11] yes, my samba repo [17:11] ahasenack: i think we can allow for building from dirty trees, i just was being overly cautious [17:12] ahasenack: ok, one moment [17:12] grrrrr [17:12] ahasenack: fakeroot is an update-alternatives postinst [17:13] ahasenack: so i'll need to fake that in my yaml :/ [17:14] /usr/bin/fakeroot-sysv [17:14] interesting [17:14] yeah [17:24] hi guys is it standard that apparmor si running and installed by default on any stock ubuntu server.? [17:24] nacc: you don't get the symlink when you install fakeroot in the snap? [17:25] ruben23: very much so, not all of the server packages have apparmor profiles, but many do [17:25] nacc: the postinst doesn't run? [17:26] nacc: you can probably pass -r and use fakeroot-sysv then? [17:27] ahasenack: yes, re: symlink [17:27] coz i wanted to move the directory of data storage for mysql database to a newly installed separate drive. and most of teh how to i see need to configure apparmor to be successfull on that [17:27] ahasenack: yes, re: postinst (by design in stage-packages) [17:28] ahasenack: i'm just making our snaps' fakeroot be fakeroot-sysv [17:28] ruben23: you can adjust the apparmor profile for mysql, in /etc/apparmor.d/local/ [17:28] ahasenack: to do what you're saying, i'd actually need to write a wrapper for dpkg-buildpackage (i'm trying to avoid that) [17:29] ah, you don't call that directly, but something else? Like debuild? [17:29] ahasenack: no, we do [17:29] ahasenack: we call dpkg-buildpackage directly [17:30] ahasenack: but it's wrong for us to call dpkg-buildpackage -r /usr/bin/fakeroot-sysv on a non-snap install [17:30] ahasenack: we should use what the user provides [17:30] ah, the run-from-tree case [17:30] ahasenack: yeah [17:30] fair enough [17:30] i guess 'wrong' is too strong [17:30] but tweaks should be in the snap, imo [17:31] "unexpected" [17:31] yeah [17:32] nacc: so to do the iso installer thing i need to do it from ubuntu and not win, correct? [17:32] madLyfe: i suppose you can do it from windows, but i have no idea how [17:33] madLyfe: seems easier to do it from linux [17:33] i found this: https://askubuntu.com/questions/409607/how-to-create-a-customized-ubuntu-server-iso not sure if it is what you were talking about. [17:33] can it be desktop? [17:33] madLyfe: to make the image? yeah [17:33] madLyfe: you just want to use an server iso to start [17:33] ahasenack: i've pusehd the fix, but the snap builders are loaded down, it seems === JanC is now known as Guest2554 === JanC_ is now known as JanC [17:34] nacc: ok, will check periodically [17:35] ok ill get setup here. [17:35] ahasenack: in this particular case, the 'success' will be that pristine-tar will fail and then we'll failover to using launchpad (note you can also pass --no-pristine-tar to `git ubuntu build{,-source}` and it should work [17:35] ahasenack: and the implication is samba also needs a reimport [17:36] what will the cron job use? [17:36] ahasenack: which cron job? [17:36] pristine-tar for new packages? [17:36] the one that imports packages [17:36] or unrelated [17:36] ahasenack: unrelated, i think [17:36] ahasenack: or maybe i'm confused [17:36] ahasenack: the importer was fixed to use the correct paths for imports as part of the pristine-tar changes [17:36] it's just because you mentioned reimport that I wondered about the cron job, because it does imports [17:37] ahasenack: that led to some breakage for old imports [17:37] effectively, we changed the hashes [17:37] ahasenack: the cron job always just imports [17:37] ahasenack: reimport is a special mode when we need to wipe a repo [17:40] tomreyn: hey, what is the scenario in that mdadm bug on xenial? Just install with raid and reboot? [17:40] does it have to be raid5, or a raid with more than two disks? [17:43] i can do this from desktop live usb, nacc ? [17:44] madLyfe: i guess so, yeah [17:45] do i need to download an new server iso or use the one that has already been put onto a usb key? [17:45] madLyfe: you can use the same iso, but you're going to end up putting that on the usb key eventually [17:46] ill just download a new one then if im going to overwrite it anyways [17:47] hmm i wonder if 17.04 would work out of the box. [17:47] is it even worth it to try that before i delve into this iso alteration? [17:51] im going to try JIC [18:00] this is 17.4 and lsusb from installer shows the USB Ethernet adapter (realtek) https://usercontent.irccloud-cdn.com/file/yah8cP1r/irccloudcapture2130928741.jpg [18:01] but not here https://usercontent.irccloud-cdn.com/file/Vd2MF8Ry/irccloudcapture870703048.jpg [18:01] during network detection [18:04] ahasenack: i'm on xubuntu 16.04.3 with mainline kernel and raid 1 with dm-crypt-luks + lvm on top [18:04] ...and i got those [18:04] running 4.12.10-041210-generic [18:05] i can try the default image if it helps - let me know [18:08] tomreyn: so luks on the md device? [18:08] or luks on each disk and then raid? [18:15] nacc: I can confirm that on the reimport my MPs seem still to match [18:16] nacc: should i also download desktop since i know thats where the usb ethernet adapter works properly? to get the correct files from? [18:16] ahasenack: ahasenack luks on top of md on top of partitions [18:17] tomreyn: ok [18:17] nacc: new MPs in the trello card and on active reviews [18:17] * cpaelzer is actually not here [18:19] cpaelzer: thanks and sorry abou tthat [18:19] madLyfe: you should just need to get the -extra .deb corresponding to the installer kernel [18:20] nacc: not a problem [18:20] nacc: if there wouldn't be issues the tool would be unused [18:20] so be happy that we find some [18:20] ahasenack: can you try your build again (snap r213/214) [18:20] * cpaelzer leaves to EOW [18:20] cpaelzer: yeah, i told you the wrong hiddne flag (or no tell) -- it's --reimport [18:20] cpaelzer: enjoy your w/e [18:20] * ahasenack fetches r214 [18:21] nacc: worked, and was very silent, only printing out this bit: [18:21] 09/01/2017 15:21:08 - INFO:Successfully built using fetch_orig_from_parent_dir(source=None) [18:21] that's a debugging detail? [18:21] ahasenack: ah you already had the tarballs in the parent dir? [18:21] probably, let me clean that up [18:21] ok, from scratch now [18:22] ahasenack: yeah, if you could clear those out and retry, it'd help [18:22] dirty tree, cleaning up [18:22] :) [18:22] ahasenack: i have a fix for that bit queued (i filed a bug to track it), it needs some thinking [18:22] nacc: exactly same thing, just that info line [18:22] ahasenack: as even with a dirty-tree, we should be able to use the pristine-tar [18:22] ahasenack: then parent_dir ? [18:23] ahasenack: then the orig tarballs were found in the parent directory [18:23] it has the source package, tarball, all the stuff [18:23] well thats confusing, if i never had internet, and the installer doesnt recognize the usb ethernet adapter at install, but can be manually added after install, it means the software is there already. [18:23] ahasenack: right, you need to delete the tarballs from the parent dir [18:23] ahasenack: in order to not use them [18:23] I did [18:23] you are lagging :) [18:24] ok wait [18:24] delete orig tarball from parent dir [18:24] cd samba; git ubuntu build-source [18:24] that should not say fetch_orig_from_parent_dir [18:24] ahasenack: are you sure you didn't misread the messag? [18:24] nacc: http://pastebin.ubuntu.com/25446444/ [18:24] ahasenack: i just tested locally and pristine-tar worked [18:25] fetch_orig_from_cache [18:25] ahasenack: *different* message [18:25] oh, would you look at that [18:25] ahasenack: clear the cache (rm -rf .git/git-ubuntu-cache) [18:26] I don't have a /home/andreas/.git/git-ubuntu-cache [18:26] in samba [18:26] ahasenack: in the git repo, i mean [18:26] ok [18:26] gone [18:26] nothing in parent [18:26] cache gone too [18:26] try the build-source again (might need a `git clean -fd`) [18:26] debian/files is always there to make me smile [18:27] yeah, that needs the build changes to use lxd to work (as we need to run clean) [18:27] ok, now it's ugly [18:27] still running [18:27] ahasenack: in your current local state, i expect you'll see pristine-tar try, fail, and then fetch_from_launchpad succeed [18:28] nacc: correct: http://pastebin.ubuntu.com/25446453/ [18:28] ahasenack: now what's interesting is that on my bastion, it built with pristine-tar [18:28] ahasenack: have you run `git fetch pkg`? [18:29] nacc: not just now, but earlier today sure [18:29] ahasenack: and/or, can you try with a fresh clone somewhere (e.g., in /tmp) and see if you also see it? [18:29] sure [18:30] ahasenack: it doesn't make sense for us to see differing behavior there -- I get (http://paste.ubuntu.com/25446470/) [18:31] Receiving objects: 93% (74038/79039), 146.04 MiB | 2.70 MiB/s [18:31] samba is big [18:31] ahasenack: :) [18:31] next time let's try base-files [18:31] ok, clone complete [18:31] switch to my branch, or try straight from ubuntu/devel? [18:32] ahasenack: your branch [18:33] nacc: http://pastebin.ubuntu.com/25446481/ [18:34] ahasenack: yeah [18:34] ahasenack: so something (seems) wonky with your local repo [18:34] ahasenack: i'm not sure why yet [18:35] ahasenack: as the last paste shows pristine-tar working [18:35] ahasenack: now, it doesn't really matter, but you do want to use pristine-tar (it's faster at least) if you can [18:37] what about this? http://fai-project.org [18:38] madLyfe: a channel regular (used to?) really like it; I've never tried it myself [18:39] dont think it will work if i dont have network though. [18:41] that's my understanding [18:42] if you don't have a network but still want customized automated installs then you have to figure out either debian installer preseeding (which is .. poorly documented at best) or figure out the kickstart compatability stuff; or figure out how to use a cloud image on the hardware in question with cloud-init [18:43] its frustrating since i dont really know nix and trying to communicate the language. [18:44] and when i see it working fine on desktop but not server. [18:56] tomreyn: ok, got it on xenial [18:56] just raid1 [18:56] tomreyn: http://pastebin.ubuntu.com/25446612/ [18:56] one per disk I'm guessing, or partition [18:57] tomreyn: is it just an annoyance? [20:59] nacc: around? [21:02] nacc: something related to manpages broke in some recent version of the git-ubuntu snap [21:03] $ git ubuntu --help [21:03] man: error while loading shared libraries: libmandb-2.7.6.1.so: cannot open shared object file: No such file or directory [21:03] same for the rest of the git (non-ubuntu) commands [21:31] ahasenack: ok, will debug, it' sprobably more PATH stuff [21:32] ahasenack: so `git tag --help` also is broken? [21:45] rbasak: cpaelzer: fyi, running the cron job from outside the snap is working. so I'm going to just let that run (it's running in verbose mode for now in a screen on my bastion) [22:58] rbasak: i'm thinking maybe we do an integration test that's just a few triival commands ... e.g., a from-scratch (no-push) import of srcpkg (maybe one without a long history, so it's fast), a git ubuntu clone, and maybe a build right after clone? [22:58] rbasak: and add that to CI? [22:58] it will slow down CI quite a bit, but I think is also our best sanity check [23:00] ahasenack: i think i agree with you as well, i will stop letting the snap auto-build propoate directly to stable [23:00] ahasenack: i *think* it's stable right now :) [23:00] let's talk it over on tuesday [23:42] powersj: --^ how hard would that be to add? (basically, if we're doing a local snap build, install that snap locally (it'd need `snap install --dangerous --classic /path/to/snap` and then a `git ubuntu import --no-fetch --no-push ipsec-tools` or something [23:44] rbasak: ooh, well, refactored (contentfully a rewrite, just due to indentation changes) of importer.py and it's working (in local testing). So that's a good sign :) [23:44] nacc: not hard at all [23:45] Can you file a bug or make a card and I can take care of it Monday [23:45] (And assign me) [23:45] powersj: good to know, i think we'll want to talk it over tuesday AM and then I'll file it then [23:45] powersj: yep, thanks -- we'll make sure the testcase is well-defined first. I think 3 simple ones would be a good first integration checkpoint. [23:46] powersj: btw us federal holiday monday [23:46] Oh yeah I get the day off! ;) [23:46] powersj: yeah :) [23:51] ahasenack: oh and `git ubuntu --help` works again [23:54] ahasenack: and just pushed a fix to silence the 'failed to find manpage for git-tag' from `git tag --help` with our gitconfig workaround