IRC logs for #baserock for Thursday, 2015-04-30

*** zoli__ has joined #baserock04:44
*** petefoth has joined #baserock05:30
*** zoli__ has quit IRC06:09
*** zoli__ has joined #baserock06:33
*** ssam2 has joined #baserock06:35
*** ChanServ sets mode: +v ssam206:35
*** mariaderidder has joined #baserock06:46
*** mike has joined #baserock06:53
*** mike is now known as Guest4598706:53
*** a1exhughe5 has joined #baserock07:06
*** zoli__ has quit IRC07:10
*** Albert_ has joined #baserock07:11
*** jonathanmaw has joined #baserock08:11
*** franred has joined #baserock08:11
franredmason in red because qemu ... probably the gcc update has caused this... -- testing now with --disable-werror08:17
richard_mawjjardon: biff re install-files symlink08:23
*** inara` has quit IRC08:30
pedroalvarezI was going to point that out08:34
pedroalvarezare there newer vrersions?08:34
richard_mawpresumably, but it's less work to fix -Werror than re-validate the new version for what we need it for08:35
jjardonpedroalvarez: yes, 2.3 was released some days ago08:36
jjardonpedroalvarez: thanks for fixing btrfs-progs BTW!08:39
pedroalvarezjjardon: as i said, it was ok :/08:41
*** inara has joined #baserock08:50
*** gary_perkins has joined #baserock08:50
jjardonrichard_maw: thanks for the review08:52
jjardontiagogomes: FYI
tiagogomesjjardon I am looking at it at the moment09:03
*** flatmush has joined #baserock09:11
*** wschaller has joined #baserock09:13
*** Krin has joined #baserock09:15
*** lachlanmackenzie has joined #baserock09:16
*** wschaller_ has joined #baserock09:27
*** wschaller has quit IRC09:30
*** flatmush1 has joined #baserock09:37
*** flatmush has quit IRC09:37
*** flatmush1 is now known as flatmush09:38
jjardonI heard there are plans for a masonv2: where is the code for that located? or its what is currently in definitions?09:55
tlsaSotK started it, and I did some fixes09:56
tlsait's on baserock-dev09:57
tlsaa gerrit driven thing that uses zuul & turbohipster, and results in a +1 or -1 depending on whether tests pass09:57
jjardontlsa: cool!09:58
jjardontlsa: so all the patches are in the mailing list? No branches somewhere?09:59
* jjardon search the archives09:59
*** flatmush has quit IRC09:59
*** flatmush has joined #baserock10:00
tlsajjardon: the're on somewhere too10:00
jjardontlsa:  thanks!10:04
tlsaa bit old now, and probably need work to merge cleanly with master10:04
radiofree  MODPOST 2759 modules10:09
radiofreeso many modules10:09
Kinnisonradiofree: cripes10:10
jjardonlooks quite cool:
franred  --> patch to fix openstack system and defo mason10:21
*** zoli__ has joined #baserock10:29
jjardonfranred: merged (FYI, 2.3.0 builds fine)10:39
franredjjardon, that change will need to test that openstack does not fail on run time ;-)10:40
jjardonfranred: yes10:42
pedroalvarezwell, we should test that anyway with the new gcc10:44
pedroalvarezit is unlikely, but for example, last week we discovered that the nfs server wasn't working since the upgrade to gcc 4.910:45
franredpedroalvarez, I don't expect any change - and I already built it10:45
pedroalvareznote, it was building, but not working10:46
radiofreethat's why it wasn't booting...11:03
radiofreewow this t510 has an nvidia11:04
rjekradiofree: That machine is monsterous; meant for people doing lots and lots and lots of kernel builds :)11:05
radiofreeor lots and lots of baserock builds!11:06
CTtpollardthe nvidia gpu in is it probably superseeded by Intel HD now11:12
radiofreehi, are there any guides for the installer?11:28
radiofreeor is it?11:28
jjardonthat page doesnt seems very generic: Id move it to a subpage and create an index there11:36
radiofreewell i'm not using the br installer yet, i want to try it out11:37
ssam2for systemd-networkd, if I want to override, should I make a file called or ?11:40
ssam2seems didn't override but I'm starting to think systemd-networkd just hates me11:45
ssam2ok, worked, but it did the thing where it kept the DHCP'ed address as well, so i rebooted11:47
ssam2I've just realised that installing .network files using install-files is going to be a pain in the ass, because there needs to be a different .network file for each system, and thus a separate directory, and a separate manifest11:51
tlsawith the baserock inport tool, is it intended that the generated lorries and stratum actually get committed to upstream local-config/lorries.git and definitions.git ?11:52
tlsaalso, has anyone looked at importing CPAN/perl?11:53
ssam2depends what you mean by "upstream" -- it's intended that they get committed to those repos in whatever Trove you care about11:53
ssam2I don't think anybody's looked at CPAN, it would be really good to have support for that11:54
tlsaby upstream, I mean on baserock.org11:55
ssam2ok. if you're importing stuff into upstream Baserock, then yes11:55
ssam2although the stratum at least will need tweaking, the generated one is *far* from perfect11:55
ssam2the generated lorries should not be too bad11:55
ssam2ideally the import tool would take account of what stuff is already in definitions, but it's not smart enough yet11:56
ssam2so you also need to resolve that manually (or fix the tool)11:56
tlsaIt wasn't clear from what I read whether it was for making updating definitions etc easier, or whether it was to avoid needing to commiting this stuff to definitions11:57
tlsabut I've got it now :)11:57
radiofreei'm booing into /dev/sda1 using a usb stick, however when i boot in mount is telling me it's "/dev/sdb" that's mounted / (it's not...)11:59
radiofreefdisk is telling me /dev/sda1 /dev/sda2 are the partitions, and it *has* booted to /dev/sda111:59
radiofreehowever if i try and mount /dev/sdb (the usb stick), it actually is mounting /dev/sda112:00
*** wschaller_ has quit IRC12:00
radiofreeif i reboot and leave the usb stick in, mount /dev/sda1 mounts the usb stick...12:00
pedroalvarezssam2: can't you configure the network using simple-network?12:01
pedroalvarezssam2: if not, you can install a generic .network file using the install-files extension, with a jinja2 placeholder12:01
pedroalvarezand then, from different clusters, you set this variable with different values12:02
ssam2pedroalvarez: oh! I forgot about the 'template' thing12:07
ssam2that's awesome12:07
ssam2simple-network.configure doesn't work, not sure why. Seems like it doesn't run at all because has not been renamed12:07
ssam2maybe it's running the wrong version, because we have to run a Morph checkout from Git to build 15.17-rc on 14.46 ...12:08
pedroalvarezI've seen people using it recently, but not myself12:08
ssam2not sure if there's any way to find out from the logs what is going on12:08
*** mauricemoss_ has joined #baserock12:10
*** wschaller has joined #baserock12:13
ssam2anyone seen 'icu' failing to build on ARM with 'error while loading shared libraries: ../../lib/ internal error' ?12:24
ssam2or, seen 'icu' succeeding to build on ARM?12:24
ssam2am going to try this patch:
pedroalvarezradiofree: ^^12:42
*** wschaller_ has joined #baserock13:06
*** ssam2 has quit IRC13:09
*** wschaller has quit IRC13:10
KinnisonDear everyone who has unmerged unabandoned patches in gerrit,13:22
KinnisonPlease sort them out13:22
KinnisonNo love,13:22
KinnisonMore usefully -- we're now at 120+ unmerged/unabandoned patches in the gerrit13:23
Kinnisonit's getting harder and harder to find things to usefully comment on13:23
KinnisonIf you have patches present which are no longer relevant, please abandon them13:24
Kinnisonif you have patches in there which just need a +2 (i.e. they have some +1s already) please chase someone to +2 them13:24
KinnisonIf you have patches with merge conflicts preventing their review/whatever, sort them out13:25
KinnisonI could start chasing people specifically, but I'd love to see us get better at self-policing our patch series13:25
Kinnisonafter all, why submit a series for merge if you're not prepared to see it through to merge?13:25
jmacsPeople aren't able to do Baserock full time, and get pulled onto other projects13:26
bashrc_looks like firehose patches are about 50% reviewed13:29
bashrc_I'm not on that project anymore and don't know the current status, so my patch series may be obsolete13:31
pedroalvarezI guess the situation would be better with more reviewers13:49
* pedroalvarez goes to do some reviews13:49
Kinnisonpedroalvarez: I've been trying to find things to review13:50
Kinnisonpedroalvarez: and it's tough13:50
pedroalvarezbut, think about the same number of patches but in the ML13:51
pedroalvarezit could be worse13:51
KinnisonIn general I find it easier to operate a mail client than a web browser :)13:51
KinnisonIt could indeed be worse13:51
pedroalvareza possible suggestion, filter the patches by project13:52
*** flatmush has quit IRC13:54
*** flatmush1 has joined #baserock13:54
pedroalvarezKinnison: you might like gertty13:56
Kinnisonpedroalvarez: perhaps13:56
radiofreessam2: i thought i fixed that :\13:59
radiofreepedroalvarez: so this installer thing14:01
radiofreehow does it work?14:01
pedroalvarezradiofree: there is a installer system, that includes the installer script on it.14:01
pedroalvarezradiofree: this installer script has to be used as 'init' in the kernel args14:02
radiofreehow do you know what root= to set?14:02
pedroalvarezif you don't know that, you have configure the installer system with initramfs14:02
radiofreei manually edited extlinux.conf to /dev/sdb to boot from a usb stick14:02
pedroalvarezwrong wrong!14:03
radiofreebut it's probably not going to be /dev/sdb on every system14:03
pedroalvarezthat's why we have initramfs14:03
pedroalvarezwhat system do you want to install?14:03
radiofreea devel system14:03
radiofreei'm going to do it the jetson way for now, e.g have a /boot/ partition and / and /src14:03
pedroalvarezok, I'm going to prepare a cluster, and then try to explain it to you14:03
radiofreesince me dding a rawdisk to /dev/sda1 didn't write the bootloader correctly14:04
pedroalvarezradiofree: oh, so you want to write the system in  a partition?14:04
pedroalvarez(all of this is to install it on a laptop, right?)14:04
radiofreeyeah, boot installer from usb, install to laptop14:05
radiofreeit actually works fine once you've booted in there14:05
pedroalvarezradiofree: how will the laptop boot the partition?14:07
pedroalvarezdon't you need anything in /dev/sda?14:07
radiofreecurrently i'm booting /dev/sda1 from a usb stick, however this obviously isn't what i want14:08
radiofreei'm not sure how to install extlinux to the mbr14:08
jjardonpedroalvarez: any idea how difficult would be make it run from the USB directly?14:11
pedroalvarezjjardon: just plug the usb, and boot from there without installing?14:12
pedroalvarezwe have been doing that for ages :)14:12
jjardonpedroalvarez: I have done something wrong here then, because it doesnt boot: I've dd the weston image (from the ci cluster) in a USB but it fails. Did I miss something?14:14
pedroalvarezjjardon: that image has in its extlinux.conf /dev/sda hardcoded14:15
pedroalvarezyour usb probably isn't there14:15
pedroalvarezI guess you see a  kernel panic14:16
radiofreejjardon: change that to /dev/sdb14:16
pedroalvarezor add initramfs14:16
radiofreepedroalvarez: so the order is dd to usb stick, that installer uses initramfs, starts the install script?14:16
pedroalvareza bit difficutl to explain :/14:17
pedroalvarezyou are trying to achieve different things btw14:17
radiofreewhat were you trying to achieve?14:17
radiofreedo you want to pop over?14:17
pedroalvareznono, you an jjardon14:18
pedroalvarezlet me paste here an example for you14:18
pedroalvarez1 minute14:18
pedroalvarezradiofree: great, we have an example in definitions:  clusters/installer-build-system-x86_64.morph14:21
pedroalvarezoh, and for jjardon this should be a good example clusters/initramfs-test.morph14:34
*** mariaderidder has quit IRC14:35
jjardonthanks pedroalvarez! This is exciting!14:39
pedroalvarezVideo of the day: Installing OpenStack on hardware using morph
jonathanmawpedroalvarez: 38 minutes is quite a long video14:42
pedroalvarezit is epic14:45
pedroalvareztiagogomes: I think I spotted a error in  51114:49
tiagogomespedroalvarez I don't think it is an error14:52
pedroalvarezdon't you have now to specify where you don't want glance?14:52
pedroalvarezI may be missunderstanding14:53
tiagogomespedroalvarez see my reply14:56
tiagogomesI will login in the compute and check, to be 100%14:56
pedroalvareztiagogomes: another thing spotted in 51214:59
tiagogomespedroalvarez confirmed, glance and keystone don't run in the compute14:59
pedroalvareztiagogomes: I changed my vote14:59
* pedroalvarez is reviewing the whole patch-set15:00
pedroalvarezoh, I see, I was wrong in 51215:01
straycatgood video but needs cats15:01
pedroalvarezstraycat: heh15:01
pedroalvarezI think that check_bool() is a confusing name15:02
pedroalvarezit checks and sets15:02
straycatlike test_and_set() !?15:06
pedroalvarezis not a big deal, but it was confusing for me15:08
pedroalvarezwith more context:
*** mariaderidder has joined #baserock15:10
pedroalvareztiagogomes: thanks for standarizing the configure extensions15:12
tiagogomespedroalvarez yw15:15
franredtiagogomes, does this implies that the setup scripts do not run in every boot?15:17
franredI though it was one thing we wanted to do but I can be wrong, pedroalvarez ^^ ?15:17
tiagogomesfranred they will run by default because they are enabled in the configuration extensions and never disabled in the configuration15:19
tiagogomesfranred, what the commit makes possible is that if you e.g. `systemctl disable postgres-setup`, it will not run anymore15:20
franredthen if you remove the dependency does not that implies that the service can run before the config-setup service?15:20
jjardonpedroalvarez: it boots but I got a kernel panic
tiagogomesbut you will have to manually disable it15:20
tiagogomesfranred for the subsequent boots? But in that case the service should already be configured and it does'nt matter15:21
pedroalvarezjjardon: oops15:22
franredin that case it would be better to disable the unit if they are not required15:22
franredtiagogomes, ^^15:22
pedroalvarezfranred: what do you mean?15:23
franredI though also that the config-setup can change values on subsequent boots if it is required... that was the reason to keep the units running after reboot15:23
tiagogomesfranred maybe but not by us, by the admin if he wanted15:23
pedroalvarezfranred: that's true, but not sure if everything is re-configurable15:24
franredpedroalvarez, yeah, I think the same...probably some values would be in the db after the first configuration15:24
pedroalvarezjjardon: I don't know how can we debug that15:24
pedroalvarezjjardon: is that master of definitions?15:24
jjardonpedroalvarez: yep15:25
pedroalvarezjjardon: you can also plug the rawdisk image to a VM as a USB I think15:25
pedroalvarez(to test what is going on)15:25
jjardonyeah, I will try to build a more generic kernel as well15:26
*** ssam2 has joined #baserock15:31
*** ChanServ sets mode: +v ssam215:31
franredtiagogomes, why postgres.service would run before postgres-setup.service?15:36
tiagogomesfranred, because it is currently on the Wants for openstack-ceilometer-db-setup.service15:38
franredtiagogomes, then postgres.service should Wants and After postgres-setup.service or remove Wants from ceilometer-db-setup.service as you have done in other services?15:40
ssam2found why simple-network.configure didn't work :) it's not listed for build-system systems15:40
pedroalvarezssam2: hah15:41
pedroalvarezsorry for not suggest that15:41
tiagogomesfranred as I stated before I prefer an approach to not pull in setup services in the Wants section. So that it makes it possible to disable by the sysadmin15:42
pedroalvarezalso, postgres can't "wants" postgres-setup15:43
pedroalvarezit won't work15:43
tiagogomesnot like things are at the moment, it would result in a loop15:43
pedroalvarezit is like keystone and keystone-setup15:43
radiofreepedroalvarez: nice video! sound!15:43
tiagogomesas the postgres.yml starts and enables postgres15:43
franredohh, sorry, I haven't noticed that...15:44
pedroalvarezI might be wrong though15:44
franredthen we should remove the Wants from the openstack-ceilometer-db-setup.service :)15:44
ssam2radiofree: thanks!15:45
franredpedroalvarez, you are not wrong, as tiagogomes said, postgres.yml (postgres-setup.service) enables and starts postgres.service15:45
pedroalvareztiagogomes: we could probably enable postgres if it's a controller node, and then put it After postgres-setup15:46
pedroalvarezmeh, I don't know15:46
pedroalvareztoo many services :)15:46
tiagogomespedroalvarez that would work15:46
tiagogomespedroalvarez, but that is orthogonal to the patch15:47
pedroalvarezI wasn't evaluating the patch on this discussion :)15:47
pedroalvarezthe patch, so far, looks ok to me15:47
radiofreeaha goot this booting from /dev/sda2, you need to install the mbr  `dd bs=440 count=1 conv=notrunc if=/usr/share/syslinux/mbr/mbr.bin of=/dev/sda`15:50
paulsherwoodwin 2415:51
edcraggi'm getting 'ERROR: Couldn't find morphology: systems/base-system-armv8l64.morph' for morph build, when the morphology is clearly there...?15:53
tiagogomesedcragg is the morphology being tracked by git?15:53
edcraggwell, it wasn't at first, and now it is15:54
tiagogomesedcragg, `git add systems/base-system-armv8l64.morph` should do the job15:54
radiofreedouble check name: matches15:54
franredtiagogomes, patch reviewed completely, thanks for clean up the system!! :)15:55
edcraggtiagogomes, radiofree: it's in upstream, and is definitely there15:55
*** a1exhughe5 has quit IRC15:55
edcraggah, name:, is see15:55
tiagogomesfranred np :) I am not done yet15:56
edcragg'name:' does match15:56
franredpdar, I have had a look to your patches in gerrit, I have some comments about them15:58
pdarcool, ta franred15:59
pdarIll have a look15:59
pedroalvareztiagogomes: re 524. I think it neutron-db-setup needs some network to run16:02
pedroalvareztiagogomes: the dependency is going to be implicit anyway, given that it depends on keystone-setup and postgres-server-setup16:03
tiagogomespedroalvarez looking at the ansible script, I can't image it why16:03
pedroalvareztiagogomes: my understanding is that to talk to keystone and postgres you need network16:03
pedroalvarezbut as I said, the dependency on network-online is going to be implicit16:04
tiagogomespedroalvarez that would make sense, so either (a) we let it be implicit, or we add network to the setup units16:05
pedroalvareztiagogomes: not sure, I remember that we had problems with this, that's why they are all over the place16:07
pedroalvarezalso we added it in "Wants" to those that have it in "After"16:07
*** zoli__ has quit IRC16:08
pedroalvarezI'd say, make it explicit16:08
*** jonathanmaw has quit IRC16:09
jjardonpedroalvarez: it worked!!!!16:10
jjardonRunning baserock from the USB drive:
franredjjardon, nice16:12
pedroalvarezjjardon: what was the failure?16:12
pedroalvarezah, you put a more generic kernel16:12
pedroalvarezahh.. the issue might be that your kernel didn't have drivers for your USB16:13
jjardonpedroalvarez: me being stupid; I did not build the initramfs before deploying16:13
pedroalvarezhah ok ok16:13
pedroalvareznice! :)16:13
pedroalvarezis it time for baserock on laptops then?16:14
jjardonThis is do cool! And I didn't know it was possible before! Thanks all!!16:14
jjardonpedroalvarez: definelitely16:15
*** petefoth has quit IRC16:15
edcraggtiagogomes: radiofree: it was a syntax error in the morph in the end16:17
franredOpenstack using latest Juno release
* pedroalvarez reviews faster16:20
* radiofree upgrades his thinkpad t510 over the network16:23
radiofreeto a weston-system!16:23
rjekradiofree: shiny!16:24
radiofreenext step, make it usable as a development machine :)16:24
radiofreewindow manager, webbrowser, e-mail??16:24
pedroalvarezradiofree: good work!16:24
radiofreepedroalvarez: should be easy to modify the installer to install the bootloader correctly16:25
DavePage_2015 is the year of Baserock on the desktop16:25
pedroalvarezand servers!16:25
*** mariaderidder has quit IRC16:27
jjardon3D acceleration with latest mesa working as well:
rjekShame about the wallpaper.16:30
ssam2franred: holy shit16:32
ssam2i thought that was going to be 6 months work or something!16:33
jjardonrjek: :) BTW,  in case you didn't noticed, this is with gcc 5.1 already16:33
paulsherwoodssam2: i think kilo will be rather more work? :)16:33
ssam2oh yeah, juno comes before kilo, not after16:34
rjekjjardon: \o/16:34
jjardonfranred: already?? nice! :)16:34
*** Krin has quit IRC16:39
pedroalvarezssam2: haha, yes, kilo is another story16:40
*** ssam2 has quit IRC16:46
*** Guest45987 has quit IRC16:50
pedroalvareztiagogomes: re - cinder_manage:
SotKif anyone has a chance to, could someone please take a look at and ?16:54
*** mauricemoss_ has quit IRC17:00
paulsherwoodSotK: did i hear something about a new/pending version of mason?17:01
*** bashrc_ has quit IRC17:01
*** sambishop has quit IRC17:09
*** gary_perkins has quit IRC17:10
*** petefoth has joined #baserock17:21
*** wschaller_ has quit IRC17:40
*** DavePage has joined #baserock17:41
*** Stanto_ has joined #baserock17:46
*** juergbi` has joined #baserock17:46
*** juergbi has quit IRC17:47
*** paulw has quit IRC17:47
*** Stanto has quit IRC17:47
*** DavePage_ has quit IRC17:47
*** Stanto_ is now known as Stanto17:50
*** paulw has joined #baserock17:53
richard_mawthanks to straycat I've worked out the kinks in `morph anchor`, it now does the minimal amount of pushing, uses the right push urls for the server and appropriately warns when it has had to degrade an anchor from a tag to a commit17:56
richard_mawwhich we need for now because we refrence tag objects in our definitions, we can't push tags to some repositories and git won't accept tags to where we can push17:57
richard_mawall that's left is bikeshedding on the name and complaints about insufficient documentation17:58
*** mariaderidder has joined #baserock17:58
richard_maw for the interested18:00
richard_mawI'd appreciate if straycat could lift his -218:01
*** lachlanmackenzie has quit IRC18:30
SotKpaulsherwood: yeah, I sent patches a few months ago and then tlsa sent a new version at some point which is still unreviewed on baserock-dev as far as I know18:37
SotKIts probably bitrotted and in need of a bit of work to rebase it, but idk18:37
*** mariaderidder has quit IRC18:38
*** zoli__ has joined #baserock18:52
*** zoli__ has quit IRC18:53
*** zoli__ has joined #baserock19:23
*** zoli__ has quit IRC19:59
*** juergbi` is now known as juergbi20:11
*** Albert_ has quit IRC21:11
*** zoli__ has joined #baserock23:23
*** zoli__ has quit IRC23:44

Generated by 2.15.3 by Marius Gedminas - find it at!