[00:15] Can anyone tell me why after juju bootstrapping, it is defaulting always to the same bootstrapping node? What information related to juju is cached? [00:15] I should elaborate, what information related to juju is cached/stored by MAAS or Cobbler? [00:26] nothing [00:26] they are separate layers [00:26] maas doesn't randomise node selection though [00:27] I just asked the question over on the juju side and they said it was maas? :) [00:28] doing the node choice, yes. [00:28] but not remembering that its a specific way [00:33] is node choice information cached/stored? [00:34] ok, I see what you are saying - no its not [00:34] why is the same node getting chosen each time then? [00:34] its not the first node that gets bootstrapped [00:35] that's what one would logically expect if the information isn't cached [01:06] burnbrighter: lowest mac address perhaps ? === matsubara-afk is now known as matsubara [09:41] roaksoax: I've tested your package on precise. The application loads fine, no signe of a wsgi error. [09:44] s/signe/sign/ === matsubara is now known as matsubara-lunch === matsubara-lunch is now known as matsubara === mrevell_ is now known as mrevell [13:01] roaksoax, looks like jeroen approved. [13:03] roaksoax, ping. [13:19] roaksoax: Want to talk about setup.py at some point? [13:28] smoser: pong [13:28] allenap: sure give me a few minutes [13:28] rvba: yeah.. that means something is screwed in my system then [13:29] roaksoax: seems so. I've done the test on a brand new canonistack precise instance. [13:29] rvba: ok awesome, It failed for me to on a canonicastack, but then worked so something on my system is screwing it for sure [13:29] Daviey: i'm ready when you are [13:36] rvba: ok so I had to manually ship the enums.js cause building it buildseverything first apparently and gets stuff from pypi it seems [13:40] roaksoax: I guess that's all right for now, just to get us going. allenap wants to change how we build things. [13:41] rvba: hehe yeah it is necessary for some part [13:41] allenap: ok so i'm all yours [13:41] allenap: I agree that what setup.py is doing now is good, though I personally believe it should build everything (or almost everythin) this way a makefile is not shipped [13:41] roaksoax: Okay, now I need a few minutes :) rvba, roaksoax: want to talk at 1400 UTC? [13:42] Ok. [13:42] Thanks. [13:42] sounds good :) [14:02] roaksoax, rvba: I've invited you to a hangout, https://plus.google.com/hangouts/_/30bec8fa14bb528af4c54648cfc5df6003d80c21?authuser=0&hl=en-GB [14:30] roaksoax: i have a call now [14:31] allenap: while you look into getting the enums.js being installed, I can look into installing other files I install manually on packaging [14:31] Daviey: ok, let me know when you free to talk then :) [14:31] roaksoax, ok. so i'm still confused about how your maas-celeryd.conf job works. [14:31] roaksoax: Do you mean you'll fix lp:maas for us? :) [14:31] obviously i'm an idiot. [14:31] everyone knows that. [14:32] but i do not think it is PYTHONPATH that is going to get the config found. [14:32] (grep PYTHONPATH from packaging branch shows only the maas executable. and the maas user doesn't even have a home directory, so its not something there) [14:32] rvba: can you explain us how exactly celery works and how come it ends up sourcing the config file from /usr/share/maas? [14:33] roaksoax: well, celery simply does "import celeryconfig". [14:33] Whatever is one the path gets loaded. [14:35] allenap: I'll fix setup.py to install those files I manually install from packaging, such as: txlongpoll.yaml or pserv.yaml, etc [14:35] smoser: ^^ [14:35] \o/ [14:36] s/one/on/ [14:36] rvba, 'path' as in PYTHONPATH ? [14:36] smoser: Yes [14:36] then i dont understand how that job is going to do this. [14:36] smoser: the job starts the daemon, which imports whatever is in the path [14:37] fwiw that's how we run it on a dev instance: http://paste.ubuntu.com/1066125/ [14:37] yes. and there is no special PYTHONPATH for that job [14:37] i can only think that the upsart job is busted then. [14:42] smoser: maybe so, I haven't thought of that :) [14:43] smoser: I just assumed that celeryd will import from /usr/share/maas [14:43] smoser: the same way maas imports its config form there [14:45] allenap: i.e. http://paste.ubuntu.com/1066136/ [14:45] err e.g. [14:58] roaksoax, so i was hoping that you could just explicitly pass it the path to the file to load [14:58] but it appears '--config /etc/maas/celeryconfig.py' doesnt work [14:58] so unfortunately, it seems that you will have to set PYTHONPATH to include /etc/maas [14:58] which seems stupid, to be honest [14:58] smoser: yeah, that would be much more clean, but unfortunately celeryd does not support it. [14:59] as it would then have implications on any python that runs from celeryd or subproceses [15:00] rvba, is this 'celeryconfig.py' actually a config file? [15:00] ie, would you ever expect for it to be modified by a user? [15:00] smoser: i'll include /usr/share/maas to PYTHONPATH [15:00] because if not, we shouldn't put it in /etc/ [15:00] roaksoax, but that will have the same implications. [15:00] if its not a config file, then lets not put it in /etc/ [15:00] and if it is a config file [15:01] smoser: it's a config file indeed. [15:01] rvba, really? [15:01] what would be changed? [15:01] i'm reading http://paste.ubuntu.com/1066166/ [15:01] smoser: I assure you :). As weird as it may sound. [15:01] and it doesn't seem like anything would be changed. [15:02] or, if the user had to change something, it would make sense to ake those chagnes in [15:02] 'user_maasceleryconfig' [15:02] but i admit to not understanding anything. [15:02] Yeah, same trick we used for the django settings file. [15:02] so this one is not a config file [15:02] but 'user_maasceleryconfig' is [15:02] ? [15:02] no? [15:02] Depends what you call a config file I guess. This configures how celery should work. [15:03] e.g. we want it to work with "CELERY_ACKS_LATE = True". [15:03] That's in that file. [15:03] ok. i'llbuy that. [15:03] so, generally, though, i tihnk it is a bad idea to put either /usr/share/maas or /etc/maas into PYTHONPATH [15:03] would not you agree? [15:04] I don't. /usr/share/maas contains the Django project. [15:04] so then that makes more sense. [15:04] Err. [15:04] * rvba checks. [15:05] rvba: yes, usr/share/maas contains the project [15:05] No, /usr/share/maas/maas contains the project. [15:06] rvba: usr/share/maas contains the 'maas' project :) [15:07] Ah, right. /usr/share/maas/maas *is* the project. [15:09] roaksoax, so in the upstart job, we modify the environment to contain PYTHONPATH=/usr/share/maas [15:09] you dont actually have to worry about appending to pythonpath, becauase when upstart runs the job its not going to have any [15:09] (ie, you dont have to PYTHONPATH=${PYTHONPATH:+:${PYTHONPATH}} [15:09] ) [15:19] smoser: ok ;) [15:29] smoser: however, is it possible to set pythonpath in the upstart job? [15:30] well, you could do it 2 ways: [15:30] exec env PYTHONPATH=... celeryd ... [15:30] or add [15:30] env PYTHONPATH='/usr/share/maas/' [15:30] export PYTHONPATH [15:30] ? [15:31] smoser: How did the config get in the python path? [15:31] env PYTHONPATH=.... [15:31] as part of the upstart job [15:31] Daviey, it did not. [15:31] .. i happened to discuss it with smoser earlier... i was confused aswell [15:32] smoser: so how did it work? [15:32] roaksoax, you'll see other examples in your /etc/init/*.conf of 'env' [15:32] oh. [15:32] its awesome. [15:32] Daviey, [15:32] celery doesn't fail if it doesnt find a config file [15:32] it just keeps running. [15:32] (it does warn) [15:33] http://paste.ubuntu.com/1066217/ [15:34] ahh [15:34] rvba: would it make sense to rename maas_local_settings_sample.py to maas_local_settings.py so it gets installed by setup.py? [15:35] celeryd really should have *some* way to pass path to a thing to load. [15:35] it does... [15:35] roaksoax: I've got no problem with that at all. That file is only there to be used by the package. [15:36] smoser: you can declare from the cmd line [15:36] Daviey, it takes a --config argument, yes. [15:36] smoser: I hit that too, but when I symlinked from etc/maas to /usr/share/maas, the warning was gone! that's why I assumed it picked up the config: http://pastebin.ubuntu.com/1066222/ [15:36] and then it tries to load that thing by essentially 'import ' [15:37] but if that-string is not in PYTHONPATH, it doesn't work. [15:37] (and you can't pass it a full path) [15:37] roaksoax, i don tunderstand how that could possibly work [15:37] do you? [15:39] smoser: I don't that's exactly why it seems strange to me why celery works [15:39] smoser: that';s why I assu,med it sourced that path [15:39] smoser: you are welcome to install the package from PPA and see it for yourself [15:41] rvba: ok cool then [15:41] roaksoax, you didn't see the error because it only goes to stderr [15:41] it doesn't go to the log [15:41] (interestingly) [15:42] and upstart jobs stderr is going nowwhere. [15:43] smoser: anywyas, it makes no sense to continue argueing about this cause I already applied that to the upstart job :) [15:54] rvba: https://code.launchpad.net/~andreserl/maas/maas_rename_settings_file.py/+merge/112800 please :) [15:55] Done. [15:55] :) [15:55] rvba: thank you ! [16:37] allenap: still around? [16:54] roaksoax: Not really. Is it urgent? [16:55] allenap: yes and no, but if you could quickly review https://code.launchpad.net/~andreserl/maas/maas_setup_data_files/+merge/112810 i'll be gratefull (as will allow me to continue) [16:56] roaksoax: Done, +1 :) [16:56] allenap: thanks! [16:56] :) [17:22] * roaksoax lunch === matsubara is now known as matsubara-afk [19:49] smoser: would you care to review/approve https://code.launchpad.net/~andreserl/maas/packaging_updates_bzr704/+merge/112837 please? [19:55] approved. [20:12] smoser: thanks