| tdaitx | slangasek, ftbfs fixed for bluez-tools in LP: #1489661 | 00:02 |
|---|---|---|
| ubottu | Launchpad bug 1489661 in bluez-tools (Ubuntu) "bluez-tools FTBFS on wily-proposed due to AM_LDFLAGS misuse" [Undecided,New] https://launchpad.net/bugs/1489661 | 00:02 |
| Logan | slangasek: ok, I'll test build and then sync | 00:16 |
| Logan | I'll deal with the transition fallout as well | 00:16 |
| Logan | doko: regarding https://launchpad.net/ubuntu/+source/gtkmm2.4/1:2.24.4-1.1ubuntu1 | 00:24 |
| Logan | can we sync over that, even though Debian didn't make the std change? | 00:24 |
| infinity | Odd_Bloke: Sorry, I fell ill and disappeared. | 01:05 |
| slangasek | Logan: if Debian is not making the package name change for the ABI transition, you should *not* sync over it but instead merge it with conflicts/replaces going the other direction, so that users can still sanely upgrade within wily | 01:10 |
| cyphermox | tdaitx: still around? I was looking at your bluez-tools diff | 01:11 |
| tdaitx | cyphermox, yeah | 01:11 |
| cyphermox | tdaitx: maybe just missing patch tags | 01:11 |
| tdaitx | cyphermox, I set it on LP earlier | 01:12 |
| tdaitx | cyphermox, or did I miss something else? | 01:13 |
| cyphermox | ah, sorry I mean DEP-3 tags on the patch itself: http://dep.debian.net/deps/dep3/ | 01:13 |
| cyphermox | but it's tiny and pretty obvious it's from you :) | 01:14 |
| tdaitx | cyphermox, thanks, I didn't knew about these | 01:15 |
| tdaitx | cyphermox, does any of the tools we use help on getting that done? | 01:19 |
| tdaitx | otherwise I will just write down a template so I don't forget this =) | 01:20 |
| cyphermox | tdaitx: good question, I don't know. I just add them manually; usually mostly just From:, Subject: amd Bug-Ubuntu: | 01:21 |
| tdaitx | cyphermox, ok, should I added those to the patch and submit the debdiff to debian again? what about lp? | 01:21 |
| cyphermox | I leave that up to you, if you want to submit it just to Debian I'll add them for you when I sponsor | 01:23 |
| tdaitx | cyphermox, alright, thanks for your help, please keep bringing those issues forward when you see them, lots to learn ;-) | 01:27 |
| tdaitx | cyphermox, found it: quilt header -e --dep3 | 01:35 |
| cyphermox | oh, cool | 01:35 |
| tdaitx | that inserts a template | 01:35 |
| cyphermox | thanks, I'll make note of that :) | 01:35 |
| cyphermox | figured there might be a way, but it wasn't bugging me enough to just vi it in | 01:36 |
| tdaitx | there was a "quilt header -e" citation in deb maintainer manual, when I looked at the man page I noticed that --dep3 param, easy as that | 01:39 |
| AwlsomeAlex | Hello? | 01:43 |
| cyphermox | good night! | 02:00 |
| Luke | how can I set a global prompt for all users? | 02:09 |
| hallyn | cyphermox: hey, were you going to push the new golang to wily today? | 02:30 |
| pitti | Good morning | 03:01 |
| tdaitx | slangasek, opencsg FTBFS on armhf because it cant find libGLES2... that happens because: | 03:21 |
| tdaitx | 1. it uses qmake-qt4 to create the Makefile, but qt4-x11 is being compiled with libgles2-mesa-dev in armel/armhf (against libgl1-mesa-dev and libgl1u-mesa-dev in other archs) | 03:21 |
| tdaitx | 2. it depends on glew, but glew depends on libgl1-mesa-dev and libgl1u-mesa-dev, thus the libgles2 dependency is not set anywhere | 03:21 |
| tdaitx | thus either glew has to be build using libgles2 in armel/armhf or qt4-x11 must use libgl1/libgl1u for armel/armhf | 03:22 |
| tdaitx | slangasek, ^ | 03:22 |
| tdaitx | slangasek, not sure which one is the right way to go (or if there is a third alternative) | 03:23 |
| Logan | slangasek: it is making the package name change, whereas we didn't | 04:41 |
| lotuspsychje | good morning to all | 05:26 |
| lotuspsychje | is it possible to add the broadcom firmware drivers to ubuntu-restricted-extras so users with bc chipsets can install ubuntu more easyly? | 05:26 |
| pitti | this is already done by ubuntu-drivers-common and in the installer | 05:41 |
| pitti | hardware drivers don't belong into ubuntu-restricted-extras, as these would be useless or even detrimental for users who don't have that hardware | 05:41 |
| pitti | lotuspsychje: ^ | 05:41 |
| lotuspsychje | pitti: does it need to have internet connection for ubuntu-drivers-common? | 05:42 |
| pitti | I believe we shipt the driver on the ubuntu images, so the installer grabs it from there | 05:43 |
| lotuspsychje | pitti: because many users on 14.04 complain in #ubuntu their broadcom doesnt get recognized | 05:43 |
| lotuspsychje | pitti: or has this been fixxed in 14.04.3? | 05:43 |
| pitti | not particularly, it was working in 14.04 already for at least some chips | 05:44 |
| lotuspsychje | pitti: maybe its the default broadcom driver and not the firmware thats added? | 05:44 |
| lotuspsychje | they have to do this sometimes: https://help.ubuntu.com/community/WifiDocs/Driver/bcm43xx#STA_-_No_Internet_access | 05:45 |
| pitti | but there might have been some changes to replace bcmwl with the one shipped by the upstream kernel | 05:46 |
| pitti | (no further idea I'm afraid, it's been many years since I had such a thing) | 05:46 |
| lotuspsychje | this user had no internet access on eth0, maybe thats why? | 05:47 |
| pitti | the *intention* is that the driver is on the image | 05:48 |
| lotuspsychje | pitti: ok tnx for investigating | 05:48 |
| pitti | lotuspsychje: and http://releases.ubuntu.com/14.04.3/ubuntu-14.04.3-desktop-amd64.list still does ship it | 05:49 |
| pitti | /pool/restricted/b/bcmwl/bcmwl-kernel-source_6.30.223.248+bdcom-0ubuntu0.1_amd64.deb | 05:49 |
| pitti | so that's not it | 05:49 |
| pitti | you need to enable [ ] Install non-free drivers in the installer for that | 05:49 |
| pitti | on the wiki page you cited it also just installs that deb from the cd-rom, so in principle it ships everything that's needed | 05:50 |
| lotuspsychje | pitti: didnt ntocie there was such option yet? | 05:50 |
| lotuspsychje | pitti: yeah i know the cdrom trick does it, but for most n00b users its a little hard to find | 05:51 |
| pitti | last time I installed on a laptop with broadcom, it "just happened", but since then I'm afraid I don't know off-hand what's going wrong | 05:51 |
| pitti | a bug report with installer logs would be helpful for that | 05:51 |
| lotuspsychje | pitti: if that occurs again on some user, ill push to bug it | 05:52 |
| pitti | thanks! | 05:52 |
| lotuspsychje | pitti: thank you for lighting things up | 05:52 |
| pitti | caribou: hm, I'm adjusting your rsyslog merge to 8.12.0-1; 8.9 is quite old, not even in testing any more | 06:02 |
| pitti | caribou: followed up to the bug | 06:09 |
| pitti | doko: hmm, I hinted gcc-5, but http://people.canonical.com/~ubuntu-archive/proposed-migration/update_output.txt says it renders packages uninstallable -- any recent dependency change in that version? | 07:17 |
| smb | rbasak, on ffe... have now sponsors and sru team subscribed to bug 1487538. are there specific sponsors I should ask in this special case? | 07:40 |
| ubottu | bug 1487538 in Ubuntu "[needs-packaging] [FFE] Add dpdk to wily universe" [Wishlist,New] https://launchpad.net/bugs/1487538 | 07:40 |
| rbasak | smb: thank you for pushing ahead with DPDK, and sorry I've been tied up. I need to get that email to upstream sorted out. Has the FFe been approved? Need that before sponsoring. Maybe arges can sponsor as he reviewed previously? | 07:48 |
| rbasak | smb, arges: can we sync on DPDK status later today, maybe? I'd like to go through my email draft with arges with a focus on getting it done and sent if possible. | 07:49 |
| smb | rbasak, can ask him today. somehow thought first upload and then accept it as ffe from the new queue | 07:49 |
| rbasak | I've seen a release team member complain about it when done that way round before. So I presume they want to ack first. | 07:50 |
| smb | rbasak, we can do that. I am having a dentist appt. soon but something like 1UTC onwards (or as soon as arges becomes concious) | 07:51 |
| doko | pitti, yes, it adds some breaks. so maybe I should just "binNMU" these | 07:52 |
| rbasak | smb, arges: invite sent. Feel free to move. | 07:54 |
| rbasak | (is arges up by then?) | 07:55 |
| smb | rbasak, I can live with that and I believe he should | 07:55 |
| rbasak | That's 8am for him | 07:55 |
| smb | ok, if he was like me, rather not but he might. we will see | 07:57 |
| === sitter_ is now known as sitter | ||
| === vrruiz_ is now known as rvr | ||
| === sitter_ is now known as sitter | ||
| melodie | hi | 09:34 |
| Odd_Bloke | infinity: No worries; I'm back around now. | 10:01 |
| pitti | Laney: I dumped my brain into https://wiki.ubuntu.com/ProposedMigration/AutopkgtestInfrastructure#Deployment FYI | 10:19 |
| pitti | Laney: does this sound reasonably comprehensible to you? | 10:20 |
| sarnold | pitti: nice | 10:22 |
| pitti | sarnold: thanks :) | 10:23 |
| sarnold | (i'll give it a Good Read later.. it just looked good on a quick skim :) | 10:24 |
| caribou | pitti: I'm fine with redoing the merge for 8.12.0-1 but this won't happen today. Hopefully by EOD monday, is that ok with you ? | 10:37 |
| pitti | caribou: sure! | 10:37 |
| caribou | pitti: I'll also drop the rsyslog.dmesg.upstart job | 10:38 |
| pitti | caribou: no, please don't drop the upstart job | 10:38 |
| pitti | caribou: just consider if you need the corresponding (and newly added) .service for this | 10:38 |
| caribou | pitti: oh, that's what I meant, sorry | 10:39 |
| pitti | caribou: ah, ok :) | 10:39 |
| doko | xnox, did you continue to work on boost rebuilds? | 10:56 |
| doko | tjaalton, https://launchpad.net/ubuntu/+source/openscad/2015.03-1+dfsg-2build1/+build/7844937 any reason that libgl1-mesa-swx11 is not built in wily? | 11:09 |
| tjaalton | doko: it doesn't exist, debian experimental got rid of it too recently | 11:11 |
| doko | tjaalton, could you fix openscad then? | 11:13 |
| tjaalton | perhaps | 11:13 |
| tjaalton | we haven't had swx11 for a few years | 11:13 |
| pitti | Laney: I filled in the remaining sections too, now; Please let me know if something is unclear | 11:15 |
| tjaalton | doko: can you fix llvm-3.7 to build on i386 ;) | 11:15 |
| tjaalton | opengl 4.1 for radeon depends on it, and mesa 11.0-rc migrated to it | 11:17 |
| doko | heh, I looked at it, but didn't find anything yet. builds in unstable | 11:18 |
| tjaalton | apparently it's caused because we build some i686 optimization | 11:18 |
| tjaalton | there's an upstream bug open since february :/ | 11:18 |
| * tjaalton tries to find it | 11:18 | |
| tjaalton | https://llvm.org/bugs/show_bug.cgi?id=22661 | 11:19 |
| ubottu | llvm.org bug 22661 in compiler-rt "No compiler-rt library is built when configured for i686-linux" [Normal,New] | 11:19 |
| tjaalton | should be that | 11:19 |
| doko | ohh | 11:24 |
| doko | ok, even with a patch | 11:25 |
| doko | tjaalton, https://launchpad.net/ubuntu/+source/llvm-toolchain-3.7/1:3.7~+rc4-1ubuntu1 and see the Conflicts/Replaces | 11:35 |
| tjaalton | doko: I'll worry about the backport closer to the release :) | 11:40 |
| jtaylor | doko: do you know why the bash package uses its internal malloc instead of glibc's? | 11:44 |
| jtaylor | performance? | 11:44 |
| doko | I have to dig around. does something not work? | 11:46 |
| jtaylor | yes, something LD_PRELOADS a library using posix_memalign during init, and later it then uses bash's free | 11:47 |
| jtaylor | -> crash | 11:47 |
| jtaylor | it is a somewhat strange app that can probably be fixed, but I'm curious about the reason for the internal malloc | 11:47 |
| jtaylor | it seems fedora bash uses glibc malloc as that is unaffected | 11:48 |
| jtaylor | or its because our bash is a pie executable? | 11:49 |
| doko | Riddell, looks like you're reverting / overwriting many of the changes I made during the GCC 5 uploads ... | 11:55 |
| doko | kde-runtime now b-d on boost1.55 again | 11:55 |
| === MacSlow is now known as MacSlow|lunch | ||
| Odd_Bloke | arges: cloud-init has been in {p,t,v}-proposed for 7 days now and the fixes have all been validated; would you mind migrating it? | 12:00 |
| mitya57 | pitti, I have a question about autopkgtest.ubuntu.com. When I search for "sphinx", it shows me this: http://mitya57.me/tmp/searcharea.png (amd64 marked as failed), | 12:14 |
| mitya57 | but http://autopkgtest.ubuntu.com/packages/s/sphinx/ indicates that it's ppc64el that fails, not amd64. | 12:14 |
| mitya57 | I guess it is a bug somewhere? | 12:14 |
| mitya57 | As I can see amd64 never failed (on the new server), and ppc64el never succeeded. | 12:15 |
| pitti | mitya57: ah yes, a funny display bug; mind reporting that against http://bugs.debian.org/src:debci ? then it's at the right place already | 12:21 |
| mitya57 | Ok, will report now. | 12:23 |
| mitya57 | pitti: debian #797198 | 12:39 |
| ubottu | Debian bug 797198 in debci "debci: search page displays wrong architectures as failing" [Normal,Open] http://bugs.debian.org/797198 | 12:39 |
| pitti | mitya57: thanks! I think I've seen this too | 12:40 |
| rbasak | smb, arges: mind if we postpone dpdk sync until a bit later. I'm feeling a bit under the weather. | 12:42 |
| smb | rbasak, In my case depends how long that bit is, have not heard of ar ges yet. So it might suit him to get awake | 12:44 |
| dobey | pitti: hey, how do we figure out why something in debci is running out of memory? | 13:03 |
| pitti | dobey: ah, what package does that affect? I can give it a bigger instance | 13:04 |
| * pitti pats his "big_packages" list in worker.conf | 13:04 | |
| dobey | pitti: ubuntu-system-settings-online-accounts seems to be hitting this issue | 13:04 |
| pitti | dobey: really? looks like the autopilot-gtk lib ABI mismatch to me | 13:05 |
| pitti | (which breaks a lot of tests indeed) | 13:05 |
| dobey | pitti: and that causes ENOMEM? | 13:05 |
| pitti | there's a silo that needs to be released, fallout from the g++ 5 transition | 13:05 |
| pitti | it causes this "IndexError: list index out of range" for sure | 13:06 |
| pitti | I think the ENOMEM is a red herring | 13:06 |
| pitti | at least I'll retry all these once the silo lands (/me bats eyelashes towards sil2100), my hope is that they'll all just work again | 13:07 |
| pitti | the history on http://autopkgtest.ubuntu.com/packages/u/ubuntu-system-settings-online-accounts/wily/amd64/ suggests that it worked fine before with the 2 GB of RAM that it has | 13:07 |
| dobey | and it worked fine at some point since the gcc5 transition too | 13:08 |
| hallyn | mwhudson: cyphermox: my laptop would really like to upgrade but won't get past the broken golang. Wlil the new golang hit wily today, or should i wget the 70 or so debs and dpkg -i them? | 13:17 |
| cyphermox | hallyn: oh, whether you'll get golang 1.5 final depends on mwhudson, but let me fix the issue with the package now | 13:20 |
| dobey | broken golang? | 13:21 |
| hallyn | dobey: yeah, file moved from golang-go.tools to golang-go | 13:21 |
| hallyn | broken pkg not golang itself | 13:21 |
| hallyn | cyphermox: ok, last i saw the bug it looked like mwhudson was done | 13:21 |
| hallyn | feh maybe i'll make a pkg in ppa which does the breaks/replace :) | 13:22 |
| dobey | golang-go breaks/replaces golang-golang-x-tools, which golang-go.tools depends on though; so an upgrade should just remove the latter | 13:22 |
| dobey | something does still Suggests: golang-go.tools though, and probably shouldn't | 13:23 |
| hallyn | not sure whether you're telling me what the pkg should do, or suggesting an easier way fo rme to get past the apt-get dist-upgrade failures | 13:25 |
| cyphermox | hallyn: he's just re-doing the analysis I had already done. | 13:26 |
| cyphermox | hallyn: though you could just remove golang-go.tools and golang-golang-x-tools | 13:28 |
| dobey | hallyn: what i'm saying is that for me, "apt-get dist-upgrade" just removed the golang-go.tools and golang-golang-x-tools packages | 13:30 |
| hallyn | dobey: it did? hm, so it sounds like i have something pinning those then. interesting | 13:37 |
| hallyn | (note i'm not the only one wit hthis failure, someone else opened the bug :) | 13:37 |
| hallyn | all right, just removed all the pkgs | 13:38 |
| dobey | hallyn: yeah, although i think i have -proposed enabled there too | 13:38 |
| dobey | so maybe it's fixed in proposed :) | 13:38 |
| dobey | yeah, i do have proposed enabled there | 13:39 |
| === MacSlow|lunch is now known as MacSlow | ||
| tsimonq2 | Hey, Launchpad is giving me a weird error when I try to change the affected package(+editstatus)...what is the proper way to edit the affected package? | 14:24 |
| infinity | tsimonq2: Not enough info really to know what went wrong. | 14:24 |
| infinity | tsimonq2: But often it's that you're trying to assign a project bug to an Ubuntu package, instead of an Ubuntu bug. | 14:25 |
| infinity | (Which isn't the most intuitive thing to sort out sometimes, but the URL you're on is a hint) | 14:25 |
| infinity | tsimonq2: Would help if you defined "weird error". | 14:26 |
| tsimonq2 | Sorry | 14:26 |
| pitti | dobey: armhf/ppc64el are uninstallable, but i386/amd64 succeeded again; autopilot-gtk and xpathselect landed | 14:26 |
| tsimonq2 | It gives me an error of: There is no package named 'nautilus-columns' published in Ubuntu. when in the selector, it shows up | 14:27 |
| dobey | pitti: yeah, unity8-autopilot seems to be uninstalalble on those archs, and is causing lots of "always failed" there | 14:27 |
| infinity | tsimonq2: It's not wrong. | 14:28 |
| Odd_Bloke | arges: Thanks! | 14:28 |
| infinity | tsimonq2: https://launchpad.net/ubuntu/+source/nautilus-columns | 14:28 |
| pitti | dobey: lots of valid candidates now \o/ | 14:28 |
| infinity | tsimonq2: Try assigning to a package that exists. | 14:28 |
| dobey | pitti: yeah. mir boottest is failing still though :( | 14:28 |
| tsimonq2 | infinity: Then why does it show up in the selector? | 14:29 |
| infinity | tsimonq2: Because it probably existed at one point in the distant past. | 14:30 |
| tsimonq2 | infinity: When I click Select... | 14:30 |
| tsimonq2 | Oh ok | 14:30 |
| infinity | tsimonq2: But is no longer in any current releases. | 14:30 |
| tsimonq2 | infinity: Ok, thank you for your time :) | 14:30 |
| cjwatson | tsimonq2: That's https://bugs.launchpad.net/launchpad/+bug/42298 | 14:40 |
| ubottu | Launchpad bug 42298 in Launchpad itself "package picker lists unpublished (invalid) packages" [High,Triaged] | 14:40 |
| tsimonq2 | cjwatson: Well that needs to be fixed XD | 14:41 |
| tsimonq2 | cjwatson: High priority XD | 14:41 |
| cjwatson | tsimonq2: I have some work in progress for it, but it's a non-trivial wodge of complicated performance-sensitive SQL | 14:42 |
| cjwatson | The last attempt to deal with it crashed and burned, so not going to rush it :-) | 14:42 |
| Simounet | Hi there. Do you know how I can assign a bug to one of the webupd8's package? https://bugs.launchpad.net/bugs/1489842 | 14:46 |
| ubottu | Launchpad bug 1489842 in Ubuntu "Nautilus columns fields not working on grid view" [Undecided,New] | 14:46 |
| pitti | dobey: hm, unity8-autopilot uninstallable on armhf? didn't we use to have a phone on arm or something? :-) | 14:52 |
| pitti | dobey: I'll retry the mir boottest | 14:52 |
| pitti | oh wait, I already did that; last three attempts failed | 14:53 |
| dobey | yeah, i'm not sure what's failing exactly though | 14:53 |
| dobey | but it's blocking my package from transitioning :( | 14:53 |
| leszek | hi | 14:53 |
| infinity | cjwatson: Was that fresh in your mind from recently looking at it, or are you turning into wgrant? | 14:54 |
| pitti | doko: retrying one more time; I'm quite used to overriding boottests, brittle as they are | 14:54 |
| pitti | err, dobey ^ | 14:54 |
| pitti | dobey: so if you want me to do that, I can | 14:54 |
| dobey | well i don't know anything about mir, so i don't know if it's always been failing there or not | 14:55 |
| leszek | chrisccoulson: can you please explain me how you solved this issue ? https://bugs.launchpad.net/ubuntu/+source/firefox/+bug/1471949 We try to build firefox with kde integration patches on launchpad ppa but the resulting 32bit deb crashes with the same symptoms | 14:55 |
| ubottu | Launchpad bug 1471949 in gcc-mozilla (Ubuntu Precise) "Firefox 39 crashes on startup or within a few seconds on Precise/x86" [Critical,Fix committed] | 14:55 |
| cjwatson | infinity: Open browser tab :) | 14:55 |
| cjwatson | infinity: wgrant does that in real life without mechanical assistance. | 14:56 |
| infinity | cjwatson: I've noticed. | 14:56 |
| cjwatson | Mere mortals like me have to outsource some of those brain cells. | 14:56 |
| infinity | cjwatson: I always assumed rapid computer lookup or firefox history help, until I saw him rattle off bug numbers over beer. | 14:57 |
| wgrant | It's not something to aspire to. | 14:57 |
| cjwatson | Of course over beer the chances of having one's bluff called are a *bit* less | 14:58 |
| wgrant | Heh | 14:58 |
| infinity | cjwatson: But call it, we did. | 14:58 |
| leszek | chrisccoulson: and one addition. We build for trusty. | 14:58 |
| infinity | cjwatson: This is why the smart phone was invented. | 14:58 |
| slangasek | tdaitx: glew can't be built with gles2, it's gl only; and qt4-x11 is not going to use gl1 for arm, because that means losing accelerated drivers. Option 3 is to make opencsg do something smarter than asking qmake for GL library information (because qmake). I thought robru had looked at one of these earlier, don't remember if it was opencsg or not - you might want to compare notes | 15:04 |
| slangasek | Logan: whereas we didn't> the launchpad link you posted clearly shows a package name change for gtkmm2.4 | 15:05 |
| tdaitx | slangasek, k, thanks | 15:07 |
| doko | barry, skipped: enum34 (0 <- 103) | 15:18 |
| doko | got: 249+0: a-68:a-44:a-35:i-36:p-35:p-31 | 15:18 |
| doko | * amd64: python3-pykmip, python3-zeroconf | 15:18 |
| barry | doko: as soon as zigo's update to experimental's pykmip lands, i'll sync that to wily. thought i'd already dealt with zeroconf, but i can't find it, so will deal | 15:35 |
| barry | oh, looks like git was updated but it wasn't uploaded | 15:37 |
| pitti | dobey: \o/ fourth time is the charm (mir boottest) | 15:38 |
| chrisccoulson | leszek, if you're building for trusty, don't use the gcc in trusty-updates | 15:56 |
| leszek | ok | 15:56 |
| flexiondotorg | cjwatson, Does Ubiquity have any support for fcitx? | 16:03 |
| cjwatson | flexiondotorg: I don't know | 16:03 |
| cjwatson | grep source or ask cyphermox | 16:03 |
| dobey | pitti: great. thanks | 16:26 |
| Logan | slangasek: sorry, there was a context switch | 16:27 |
| Logan | slangasek: the original one I was talking to you about was gtkglextmm, where we didn't make the name change but Debian did | 16:27 |
| Logan | but they also bumped a build dependency on gtkmm2.4 to Debian's latest version, which included all of the Ubuntu changes except for changing the std | 16:28 |
| Logan | so I was asking doko if we could just pull that in and whether the std change was necessary | 16:28 |
| cyphermox | flexiondotorg: it doesn't do anything special about it, that I know. might want to grep the source as cjwatson suggested | 16:30 |
| smoser | cyphermox, https://bugs.launchpad.net/ubuntu/+source/multipath-tools/+bug/1489959 | 16:31 |
| ubottu | Launchpad bug 1489959 in multipath-tools (Ubuntu) "power8 scsii disks get blacklisted from multipath / curtin installed system doesnt boot" [Undecided,New] | 16:31 |
| smoser | wonder if you have any thougths tehre, or would be able to debug that. | 16:31 |
| cyphermox | do you just add multipath-tools in curtin? You'd also probably need sg3-utils | 16:35 |
| cyphermox | (if it's udebs you want sg3-udeb) | 16:35 |
| smoser | if you need that, should you not depend ? | 16:36 |
| smoser | cyphermox, 'apt-get install multipath-tools-boot' is all curtin does. | 16:36 |
| smoser | so even a recommends would get pulled in | 16:36 |
| cyphermox | at which point is this then? when starting the install or after the system is installed? | 16:37 |
| smoser | well, the install detects that it needs multipath (by looking for identical block devices, not by using multipath at all) | 16:37 |
| cyphermox | sure | 16:38 |
| smoser | and then installs multipath-tools-boot if necessary , configures target and expects boot to work. | 16:38 |
| smoser | by configures, i mean : | 16:39 |
| smoser | - writes 'user_friendly_names yes' to /etc/multipath.conf | 16:39 |
| cyphermox | ok, so you need to pull in sg3-utils-udev too, and probably check whether /etc/multipath/wwids and /etc/multipath/bindings are copied from the installation into the installed system (and the initramfs is updated) | 16:39 |
| smoser | - writes /etc/multipath/bindings for the boot device | 16:40 |
| smoser | - updates initramfs. | 16:40 |
| cyphermox | in the bug there isn't enough information to tell -- if you add all of multipath -v3 I could tell | 16:40 |
| cyphermox | ah | 16:40 |
| cyphermox | yes, then it's probably the wwids file missing | 16:40 |
| smoser | i attached -v3 | 16:40 |
| smoser | i dont thinks o. | 16:40 |
| smoser | even in running environment '-ll' does nothing | 16:40 |
| smoser | when it should | 16:40 |
| cyphermox | hm, it shouldn't complain about the properties missing | 16:41 |
| cyphermox | what does multipath -t say? | 16:41 |
| doko | Logan, context? | 16:42 |
| smoser | sg3-utilshttp://paste.ubuntu.com/12215510/ | 16:42 |
| smoser | http://paste.ubuntu.com/12215510/ | 16:42 |
| Logan | doko: I was wondering if we could sync over the changes in https://launchpad.net/ubuntu/+source/gtkmm2.4/1:2.24.4-1.1ubuntu1 | 16:42 |
| Logan | or if we need to merge in the std=c++11 change (which wasn't applied in Debian) | 16:42 |
| doko | Logan, sync it. I think they reverted back to an older glibmm | 16:43 |
| smoser | cyphermox, thats on the wily system that booted without multipath , then i apt-get install multipath. it did correctly (as you implied) gety sg3-utils. but seems still to not consider those disks as multipath. when in vivid it does. | 16:43 |
| cyphermox | check if udevadm info --query=all --name=/dev/sda shows SCSI_IDENT_* properties | 16:45 |
| cyphermox | that's pretty much all there is to it -- these properties are used instead of just ID_SERIAL to decide if it should be multipath devices | 16:45 |
| smoser | http://paste.ubuntu.com/12215543/ | 16:46 |
| cyphermox | and you said sg3-utils-udev is installed? | 16:46 |
| infinity | multipath-tools depends on it, I don't see how he could be missing it. | 16:47 |
| smoser | for reference, same trusty: http://paste.ubuntu.com/12215554/ | 16:47 |
| smoser | and yes, sg3-utils-udev is installed. | 16:47 |
| infinity | But it might need some udev prodding if it's installed post-boot? | 16:47 |
| smoser | cyphermox, with vpn... can you get to 10.245.71.133 ? | 16:48 |
| cyphermox | well, the initramfs needs to be updated but postinst probably should have done it | 16:48 |
| smoser | i can let you poke at a trusty box and a wily box if you can. | 16:48 |
| smoser | initramfs is updated. | 16:48 |
| infinity | And rebooted since? | 16:48 |
| smoser | curtin does that explicitly. making 14 times on average that initramfs gets updated in an install :) | 16:48 |
| cyphermox | I think he wants to see the devices first | 16:48 |
| cyphermox | but even then, multipath -ll shouldn't say that the devices are blacklisted | 16:49 |
| infinity | Well, if we've not rebooted, updating initramfs makes no difference. :P | 16:49 |
| infinity | And adding udev rules post-boot also does nothing. | 16:49 |
| cyphermox | well, udev would need to be prodded anyway | 16:49 |
| cyphermox | smoser: I can get to it but I have no access | 16:49 |
| smoser | right. my expectation is that multipath -ll should work | 16:49 |
| smoser | cyphermox, ok. i'll let you in then | 16:50 |
| infinity | You'd have to re-trigger udev if you want the new rules to kick in. | 16:50 |
| cyphermox | yep | 16:50 |
| smoser | infinity, well, presumably that would happen on reboot. | 16:50 |
| infinity | Probably just 'udevadm trigger -s scsi' or similar. | 16:50 |
| smoser | and it does in trusty -> vivid and did in wily up until recently. | 16:50 |
| infinity | smoser: Sure, on reboot it would, which was why I was trying to clear up if this was broken after your apt-get, after reboot, or both. | 16:50 |
| infinity | smoser: Specifically, the lack of SCSI_IDENT stuff in your paste. | 16:51 |
| cyphermox | smoser: you can't compare trusty/vivid to wily when there's been a major update | 16:52 |
| smoser | cyphermox, ok. you should be able to ssh into ubuntu@10.245.71.133 (wily) and for reference ubuntu@10.245.71.134 (trusty) | 16:52 |
| smoser | both systems be nice and rebooting probably need to ask someone else about. | 16:52 |
| smoser | ie, doko is using the wily system right now. | 16:52 |
| cyphermox | ran udevadm trigger -s scsi, that's it | 16:53 |
| infinity | cyphermox: I win? | 16:54 |
| flexiondotorg | cyphermox, Thanks. I did grep the code, no mention of fcitx. I was just checking there wasn't some additional functionality tucked away in the indicators. | 16:54 |
| cyphermox | infinity: well, it's still not creating the paths, but if the devices are already loaded it might be hard | 16:55 |
| infinity | cyphermox: But that made the extra attributes show up? | 16:56 |
| smoser | yeah. it did. | 16:56 |
| cyphermox | no multipath-tools-boot installed won't help rebooting on multipath | 16:56 |
| infinity | cyphermox: If so, you could retrigger the multipath events afterwads, I suspect. | 16:56 |
| cyphermox | infinity: yep | 16:56 |
| cyphermox | the properties show up now | 16:56 |
| infinity | cyphermox: Alternately, one could manually install s-thing-udev, udevadm trigger, then install multipath-tools, which might get it in the right order to not have to do more triggering. | 16:57 |
| infinity | cyphermox: Bonus point, perhaps s-thing-udev could trigger -s scsi in its postinst, and this would Just Work? | 16:57 |
| cyphermox | infinity: yeah | 16:58 |
| cyphermox | still there are other devices on that system and they should be getting picked up | 16:58 |
| smoser | http://bazaar.launchpad.net/~curtin-dev/curtin/trunk/view/head:/curtin/commands/curthooks.py#L468 | 16:58 |
| smoser | thats what curtin does for installation of mutipath | 16:58 |
| smoser | we explicitly write multipath.conf and multipath/bindings | 16:59 |
| cyphermox | there is definitely no multipath-tools-boot or /etc/multipath/bindings on that system right now | 17:00 |
| smoser | right. because i disabled installation | 17:00 |
| smoser | because if i enable it, it doesnt boot | 17:00 |
| cyphermox | so we're currently looking at two very different issues | 17:01 |
| cyphermox | along with writing the bindings file you really should add writing a wwids file | 17:02 |
| smoser | where does that go and what would it look like ? | 17:02 |
| cyphermox | it's format is pretty much /<SCSI_IDENT_LUN_T10>/ per-device or something like it | 17:03 |
| cyphermox | it goes along with bindings in /etc/multipath/ | 17:03 |
| cyphermox | just running multipath -v0 will write both files for you though | 17:03 |
| smoser | but not on diamond ? | 17:04 |
| smoser | # multipath -v0 ; echo $? | 17:04 |
| smoser | 1 | 17:04 |
| smoser | or are you saying you need a reboot? | 17:04 |
| cyphermox | oh wait | 17:04 |
| cyphermox | multipath -W will write the wwids file for you | 17:05 |
| cyphermox | for some reason on diamond it's not liking any of the devices | 17:05 |
| smoser | :) | 17:05 |
| smoser | yes, thats the problem :) | 17:05 |
| cyphermox | meh | 17:05 |
| cyphermox | it's also not really in the best of state to be debugging this | 17:06 |
| smoser | why ? | 17:06 |
| smoser | its booted off one device, but it has 5 others that are not used. | 17:06 |
| smoser | cyphermox, so why does 'multipath -ll' not work now ? | 17:09 |
| cyphermox | Aug 28 17:10:24 | sdf can't store path info | 17:10 |
| cyphermox | ^ I don't know | 17:10 |
| smoser | ok,thank you for your help. | 17:11 |
| smoser | is it ok if i leve you to poke at that, and can you give me some suggestion of how i should install and configure multipath support ? we've not had to write a WWID file in trusty -> vivid. | 17:16 |
| smoser | it appears maybe i need to write the wwid in /etc/multipath/wwids ? which personally seems quite silly if i've declared the same id in /etc/multipath/bindings. | 17:17 |
| cyphermox | I doubt writing the wwids will help in this case, looks like a legit bug for these devices | 17:18 |
| === davmor2 is now known as davmor2_HOLS | ||
| smoser | ok. well i'll leave you to debug, ok ? | 17:19 |
| smoser | thank you again for your help. | 17:19 |
| === wendar_ is now known as wendar | ||
| slangasek | Logan: ah yes :) | 17:37 |
| Logan | doko: Bug 1489980 | 17:51 |
| ubottu | bug 1489980 in gtkmm2.4 (Ubuntu) "Sync gtkmm2.4 1:2.24.4-2 (main) from Debian unstable (main)" [Wishlist,New] https://launchpad.net/bugs/1489980 | 17:51 |
| === mwenning is now known as mwenning-rr5 | ||
| === brainwash_ is now known as brainwash | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
