IRC logs for #baserock for Monday, 2015-03-02

*** petefotheringham has quit IRC00:07
*** petefotheringham has joined #baserock00:07
*** AnxiousGarlic has joined #baserock01:36
*** AnxiousGarlic has left #baserock01:36
*** ctgriffiths has quit IRC04:47
*** ctgriffiths has joined #baserock04:48
*** zoli__ has joined #baserock06:03
*** petefoth has quit IRC06:14
*** zoli__ has quit IRC06:59
*** zoli__ has joined #baserock07:23
*** sherm_ has joined #baserock07:23
*** petefoth has joined #baserock07:26
*** a1exhughe5 has joined #baserock08:03
*** pacon has joined #baserock08:03
*** mdizzle has joined #baserock08:38
*** jonathanmaw has joined #baserock08:57
*** bashrc has joined #baserock09:05
*** straycat has joined #baserock09:13
straycatbtw, in case anyone didn't notice the security alert, you should change your freenode password if you have one09:16
straycat(it's taken me several months to notice)09:18
*** CTtpollard has joined #baserock09:18
* bashrc doesn't have a freenode password09:25
*** 92AAAPL92 has joined #baserock09:29
*** gary_perkins has joined #baserock09:39
*** ssam2 has joined #baserock09:59
*** ChanServ sets mode: +v ssam209:59
* rjek tries to decide if a security issue tracker should exist along side (or integrated with) Lorry10:18
KinnisonAll lorry does is move commits around10:21
KinnisonI'd suggest any kind of tracker should operate off the git repos directly10:21
Kinnisonperhaps by being notified by the git server when new commits show up10:21
rjekYeah10:23
rjekHmm.10:23
rjekI think what I need to do is set up Lorry on my laptop and have it do some things so I can see how they might be integrated or at least talk to each other10:24
KinnisonLorry has a non-trivial number of dependencies10:24
Kinnisonit may be easiest to set up a trove for yourself and poke at that, I believe there's documentation on doing that on the wiki10:25
rjekOK.  I'll have a look.10:25
rjekI naively thought Lorry would just be a python script that required git, cvs, svn, etc to be installed.10:25
KinnisonLorry itself yes10:28
Kinnisonif you want the controller to work, etc, then you will need more10:28
rjekRight.10:28
rjekI thought the "controller" was a static piece of HTML that it spat out :)10:28
Kinnisonlorry needs cliapp etc10:28
rjekI may be somewhat behind the times.10:28
Kinnisonthe controller is a webapp and a bunch of timer scripts etc10:28
*** gfinney has joined #baserock10:48
*** gfinney has quit IRC10:52
*** wschaller has joined #baserock11:41
kejiahuI am trying to set up a Trvoe server and distbuild network on Moonshot nodes, and all the them will netboot from a nfs server set up on another node. should I provide the system definitions and cluster files for deployment after I have verified that11:41
ssam2kejiahu: is your question whether you should send the definitions for this to be included in the baserock reference system definitions Git repo, or keep them somewhere in a different repo ?11:47
ssam2I don't really know what's best, but definitely make them public if you can even if it's in a fork of definitions.git on github or somewhere else11:47
pedroalvarezkejiahu: if the Trove is only needed for the distbuild network, I suggest you to use it only as artifact-cache-server11:48
pedroalvarezkejiahu: to achieve that just don't set any 'UPSTREAM_TROVE' in the cluster morphology of the trove11:49
kejiahussam2: the cluster will be very customised to the environment we used for the netboot, so should I provide my cluster file? or just a template or sorts of things?11:51
kejiahupedroalvarez, got it, thanks11:51
pedroalvarezkejiahu: once you finish them and you have verified that they work, please share. Maybe the baserock-ops team is interested :)11:52
kejiahupedroalvarez, yes, I will do that11:53
ssam2kejiahu: provide everything, unless it's secret :)11:54
radiofreepycrypto failing to build for me11:56
radiofree    Merge branch 'baserock/pedroalvarez/pycrypto-fix-morphology'11:56
kejiahussam2: ok11:56
radiofreewill try again11:56
pedroalvarezradiofree: yes please11:57
pedroalvarezmy mistake11:57
*** pacon has quit IRC12:10
franreddoes someone has time for a quick review of a lorry http://paste.baserock.org/vewirunune ? this is an openstack package12:11
pedroalvarezfranred: new dependency?12:12
franredpedroalvarez, no, it is an old one, but I never submit the patch for it12:12
pedroalvarez+112:14
*** gfinney has joined #baserock12:18
ssam2franred: +112:20
franredpedroalvarez, ssam2, thanks for the review! it has been merged12:30
*** wschaller has quit IRC12:31
pedroalvarezIt's horizon who needs it as a dependency12:31
straycathttp://sprunge.us/iaFB <- openstack-ansible-modules lorry patch12:44
pedroalvarez+1! I'd be happy if we add them to ansible stratum12:45
franredstraycat, they are not in the openstack repository: http://git.openstack.org/cgit/  wouldn't be better to put it in a different lorry file?12:48
straycati don't care really, i just put it in openstack because it's an openstack thing12:49
persiaI think putting it in the openstack file is fine: stuff moves between openstack and stackforge dependent on the will of the OpenStack Technical Board.12:53
persiaTrying to make our lorries match their decisions will be annoying enough if we have to change the repo locations, let alone if we have to edit two files in the process.12:54
franredfair enough, straycat +112:54
straycatmerged13:07
*** cyndis has quit IRC13:27
pedroalvarezstraycat: so my idea for those modules was something like this: http://paste.baserock.org/raw/nubofiboya13:30
*** cyndis has joined #baserock13:36
*** cyndis has quit IRC13:42
*** cyndis has joined #baserock13:44
*** CTtpollard has quit IRC14:00
*** wschaller has joined #baserock14:07
tlsadoes anyone know why artifact metadata may not be fetched from the cache server?14:08
ssam2in what situation?14:12
*** rdale has joined #baserock14:13
ssam2the code to transfer artifacts around is duplicated in several places right now and each implementation is lacking, each in a different way14:13
ssam2hopefully they will all be consolidated in the ostree branch14:13
92AAAPL92Any suggestion on the approach of extending install-files to install files from more than one manifest? The problem is finding a good separator, as everything besides the null char can be in a filepath14:14
*** edcragg has joined #baserock14:14
*** 92AAAPL92 is now known as tiagogomes14:14
* tiagogomes doesn't have a clue from whee 92AAAPL92 came from14:15
tiagogomesWe could restrict the directories in definitions which contain manifests to not have ':' in its filename14:16
*** CTtpollard has joined #baserock14:16
ssam2you could use the same trick as fstab.configure14:21
ssam2and match any environment variable that begins INSTALL_FILES14:21
ssam2so you could have INSTALL_FILES_DISTBUILD, INSTALL_FILES_MASON, or whatever14:21
ssam2personally I think we should kill install-files and allow specifying the manifest in the cluster morph directly, and in a more human-friendly format14:22
tiagogomesI asked my question without looking at install-files first, it appears it is already supported14:22
tiagogomes   manifests = shlex.split(os.environ['INSTALL_FILES'])14:22
tiagogomes        for manifest in manifests:14:22
tiagogomesI particular like the more human-friendly format14:22
* kejiahu is still interested on what is fstab.configure trick14:23
ssam2it parses all variables that start with FSTAB_14:23
ssam2so you can specify multiple fstab entries14:23
kejiahuoh, yes, thanks14:24
tiagogomeskejiahu try `INSTALL_FILES: "moonshot/manifest distbuild/manifest"`14:24
kejiahutiagogomes: will do14:26
Krinhi All, have a bit of a problem with downloading a tarball to test. i'm pulling the tarball from here: https://pypi.python.org/pypi/mox3/0.7.0 the tarball downloads and opens fine manually, but when i point morph to that location it failed to read the tar because of a 'short read error' does anyone know what this may be?14:45
pedroalvarezKrin: when you say "point morph to that location" you mean thay tou are using a tarball url in the 'ref' field?14:47
Krinin the repo feild. should it be in the ref feild then?14:47
Krinhttp://paste.baserock.org/daqoyusozo is the thing i put together with my poor morph knowledge14:47
pedroalvarezoops sorry14:47
pedroalvarezyeah, on repo field14:48
radiofreeKrin: https://pypi.python.org/pypi/mox3/0.7.0 is a webpage?14:48
pedroalvarezAlso, repo field can only point to git repositories14:48
Krinyes, with a tarball on it, which makes no sence to me but it's worked before and been explained to me as "morph magic"14:48
pedroalvarezKrin: can you do `git clone https://pypi.python.org/pypi/mox3/0.7.0` ??14:49
pedroalvarezI don't think so. looks like mox has its sources in http://pymox.googlecode.com/svn/trunk/ , which is a svn repository (not git)14:50
Krinso will i have to grab the tar manually and upload it to a repo?14:51
pedroalvarezKrin: or you can just copy the tarball somewhere in your baserock system, convert that place to a git repo, and point to that folder in your repo field14:52
radiofreeKrin: i tried it and got http://fpaste.org/192216/30795514/14:52
radiofreeso i believe your "short read" is there because morph initially tries to download a tarball containing the repo from gbo14:53
radiofreeobviously it's not there, so moves onto trying to git clone that repo url, that's where the failure is14:53
Krinok, the repo that Zara found looks good to me, so i'll try that one :)14:54
Zara( https://github.com/emonty/pymox )14:54
Krini was only trying it because we did something like that for pytz. i'v never understood why it worked >.<14:54
*** sherm_ has quit IRC14:55
radiofreeare you sure you weren't pointing it at a git repo last time?14:55
Krinpretty sure, but not 100%14:55
radiofreeyou can use http to clone git repos btw14:56
Krini know, i'm assuming it was a simmilar case because the webpage that we took pytz from looked almost identical.14:57
pedroalvarezoh yeah, the svn repo i posted is the original pymox, and not pymox314:58
pedroalvarezso that github one might be the official14:58
*** sherm_ has joined #baserock15:09
*** wdutch has joined #baserock15:14
straycatpedroalvarez, I'm not sure, there's a core keystone_user module too http://docs.ansible.com/keystone_user_module.html so I'm wondering if it's best to install the openstack-ansible-modules somewhere else and make playbooks use the openstack modules by setting the ANSIBLE_LIBRARY environment variable, ftr fedora seems to put them in /usr/share/ansible/ansible-openstack-modules15:17
pedroalvarezI was using the -M option of ansible-playbook15:21
straycatok15:21
pedroalvarezI'm not sure about the ANSIBLE_LIBRARY approach, is there a way to set a env variable?15:21
pedroalvarezof course there are ways, but I mean like adding the variable to a file which is loaded every boot15:22
straycatthe unit could run ANSIBLE_LIBRARY=... ansible-playbook ...15:23
pedroalvarezin that case I'd prefer -M I think15:24
pedroalvarez/usr/share/ansible/ansible-openstack-modules works for me15:25
straycatthe README also said you can make a 'library' symlink in the directory that contains the playbook15:25
pedroalvarezstraycat: what do you prefer?15:27
pedroalvarezwell, they don't make any difference when installing these modules, so we don't care about that right now15:29
straycat-M i think15:32
* straycat wonders whether pedroalvarez has a reworked version of http://paste.baserock.org/raw/nubofiboya he could cherry-pick :316:08
pedroalvarezoh! I though you were going to add it :)16:08
straycatoh well, was worth a try :p16:09
straycati can add it then16:09
pedroalvarezI will give you a sha116:09
pedroalvarezstraycat: you should be able to cherry pick 0d544a47edd3b688436d592a7ce2a5fa34a6377f16:12
straycat<fx:monty> excellent16:13
straycatpedroalvarez, was there a specific thing that required you to move to 1.8.4 ?16:18
pedroalvarezno, I just wanted to  have it upgraded because newer versions may behave better16:19
straycatokay16:19
mauricemoss_I was creating some confusion not pointing to my GitHub repos for the mips64b cross-bootstrap patch. Before I resend the revised version, can someone have a look at the patches its dependent on?16:24
mauricemoss_I know this adds friction and the preferred way for cpython and binutils would be to submit them upstream, but I fear upstream won't take these hardware specific patches.16:25
*** wschaller_ has joined #baserock16:45
*** Codethink__ has joined #baserock16:46
*** wschaller has quit IRC16:47
*** wschaller_ has quit IRC16:50
*** Codethink__ has quit IRC17:01
*** wschaller has joined #baserock17:02
*** jonathanmaw has quit IRC17:02
*** a1exhughe5 has quit IRC17:19
*** zoli__ has quit IRC17:23
*** gary_perkins has quit IRC17:30
*** mdizzle has quit IRC17:43
*** zoli__ has joined #baserock17:43
*** bashrc has quit IRC17:55
*** tiagogomes has quit IRC17:57
*** sherm_ has quit IRC18:02
*** ssam2 has quit IRC18:09
*** edcragg has quit IRC18:10
*** gfinney has quit IRC18:14
*** wschaller has quit IRC18:23
*** gfinney has joined #baserock18:40
*** gfinney has quit IRC18:45
persiamauricemoss_: have you sent them upstream?18:45
persiaIn the past we accepted patches that were still under review, as long as the process of upstreaming had started.18:46
*** straycat has quit IRC19:58
*** straycat has joined #baserock20:02
* straycat has a problem caused by a bug in the version of busybox grep we're currently using, it's fixed by 73ed05eea566ac90c9e690488d2acb4375d52e7520:47
straycatwe have quite a delta against busybox so not sure how folks would want to get that fix in20:50
* straycat disappears20:50
*** zoli__ has quit IRC21:19

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!