/srv/irclogs.ubuntu.com/2014/05/19/#maas.txt

=== hazmat` is now known as hazmat
=== vladk|offline is now known as vladk
=== CyberJacob|Away is now known as CyberJacob
=== vladk is now known as vladk|offline
=== vladk|offline is now known as vladk
=== CyberJacob is now known as CyberJacob|Away
jtvgmb: that file was part of the simplestreams code drop, which lacked tests.08:21
jtvWe did some of the most basic cleanup, and there's an end-to-end test, but more unit tests are still needed.08:21
bigjoolsgmb: loads of that import code is untested, we're adding it as we go08:32
bigjoolsjtv: perhaps if you have some time you might want to take a look09:02
bigjoolshttps://code.launchpad.net/~julian-edwards/maas/report-subarches-in-bootimage/+merge/21998709:02
bigjoolsat least someone please give me something I can work on tomorrow :)09:02
jtvbigjools: is this just because when I commented on it earlier it was in the wrong channel, which _you_ picked?09:03
jtvOr is this a different one?09:03
rvbagmb: btw, I just noticed that you used django's BinaryField in BootSource.  It's all right but it means we now have 2 different BinaryFields in use: the one we created (back when the other one did not exist) and the one provided by Django.09:03
bigjoolsjtv: same one09:04
jtvIf we had time, I'd love to look into abolition of our home-grown BinaryField...09:04
bigjoolsjtv: doesn't need to be you, just *someone*, please!09:04
rvbaYeah, that would be great.09:04
rvbabigjools: I'll have a look later today.09:04
jtvbigjools: as I said, needs documentation.  :)09:05
bigjoolsjtv: next branch09:05
jtvOK.  One thing that's still sadly lacking is a mention of what that meta file is _for_.09:06
jtv(And if you're going to chide me for the terminal preposition, up shut)09:06
bigjoolsjtv: an abomination up with which I shall not put09:08
jtvOn which, spot.09:08
jtvIt's awkward and off it pisses me.09:08
bigjoolsI shall pop in some comments to explain for what the file is used09:09
jtvHuh what?09:09
jtvOh, I see what you did there...09:09
jtvbigjools: I think extract_image_params suffers a bit from the vagueness of that term "params."  It makes you call individual elements of params "param" despite it containing multiple parameters.09:30
jtvIs there no more specific name you can give each of those dicts?09:30
bigjoolsjtv: yes, like I said, the branch needs some tidying09:30
bigjoolsit has evolved somewhat09:30
jtvJust adding specifics, because you needed things to work on...09:31
bigjoolsjust bear in mind this is nothing I've added :)09:31
jtvAFAICS "param" is new.09:32
bigjoolsoh that bit yeah09:32
bigjoolswell the function name started it I supposed09:33
bigjoolsextract_image_params is returning a variable called... params.  Fancy :)09:33
bigjoolsI need to EOD now, rather tired09:35
bigjoolsall that coding has left me exhausted!09:35
jtvGood night!09:35
bigjoolsthanks for looking at the branch09:35
bigjoolsI appreciate your comments09:35
jtvApart from the one about your face, obviously.09:36
jtvGood night.  :)09:36
bigjoolsyou said nice things about it once09:36
jtvYou mean the one where it was too high for me to see?  :)09:37
bigjoolsor maybe that was in one of your dreams09:37
* jtv wonders how bigjools would hear him talk in his sleep...09:37
bigjoolsI am turning this PC off now before this gets too exciting09:37
jtv*chuckle*09:37
bigjoolscheerio :)09:37
jtvnn09:38
rvbagmb: arg, it's getting worse: a BinaryField is non-editable.  Which means we can't use forms to accept base64-encoded input to change the value of that field.10:14
gmbrvba: Damn.10:16
rvbagmb: I'll figure a way… but it will require changes to the model…10:17
gmbYeah.10:17
gmbrvba: So, what about (this is horrible) storing the keyring data as a (base64-encoded) string, and using a file upload field to get the keyring data in there (and then dealing with the file behind the scenes).10:17
gmbCould that work with the API/CLI?10:17
* gmb doesn’t know whether we do file-stuff already within the CLI.10:17
rvbaI seem to remember the CLI supports file upload now…10:18
jtvIt does10:18
rvbagmb: it would be better to figure out a way that work with Django's binary field though.10:18
rvbaworks*10:19
gmbrvba: Definitely. I’ll happily defer to your Django experience rather than my hacktastic approach :).10:19
jtvgmb: BootSource.objects.get_boot_sources_for_cluster() takes more typing than doing the query yourself.  Maybe we should just have a 1:n attribute on NodeGroup.11:01
gmbjtv: Totally WFM.11:02
gmbjtv: I’ll stick a card on the board.11:02
jtvLow priority though.  :)  This is tech debt.11:02
jtvFairly easy to pay off.11:02
gmbYeah.11:03
gmbjtv: Actually, IIUI, it should be there already as NodeGroup.bootsources_set.11:03
gmbBut I’d forgotten that when I wrote the function :)11:03
jtvMight be, actually.  I forget whether that needed any work.11:04
gmbI think it’s automatic.11:04
* gmb -> Dinner (by which I mean Northern ‘dinner,’ which Southerners call “Lunch”).11:21
* jtv very confuse11:22
rvbagmb: I think I spotted another problem.  We use a FilePathField for BootSource.keyring_filename.  If you have a look at the doc you'll see that this will validate against the *region*'s file system.  Where in fact this file only has meaning in the context of the cluster.11:32
jtvI thought it was intended for use to refer to a region-side file..?11:39
jtvIIRC the region then reads that file and sends the _contents_ to the import code.11:39
rvbaThen I don't understand why we have bot a keyring_filename and a keyring_data field.11:44
rvbaboth*11:44
jtvMy understanding was: if you set a filename, it will get read from the filesystem every time the imports are run, which means the keyring could come with a package, and get automatic updates.11:46
jtvWhereas if you have a custom keyring, you can pass the keyring contents centrally and have the import code use that instead, and you manage updates yourself.11:46
jtvBut I'm only guessing, really.11:46
rvbaThat would make sense.  I'll let gmb confirm that it's what he had in mind.11:47
jtvAnd document.  :)11:47
rvbajtv: then I guess we need to think about migrating the config from the cluster's config file to the region's DB.  Because the meaning of the keyring_filename is changing (from a file on the cluster to a file on the region).11:48
jtvI thought the plan was not to migrate those files.11:49
rvbaSo, use the default?11:49
jtvYes, download the world.11:49
jtv(For Precise/Trusty).11:50
jtvWith a UI for those who want to pare it down.11:50
rvbaI see.11:50
jtvgmb: could you also confirm to us that there is no plan to migrate existing bootresources.yaml files?  I've been working on the assumption that we always start off with The World.12:01
jtv(Things get a lot more complicated otherwise.)12:01
jtvgmb: I see there's a card for removing parts of bootresources.yaml.  I'd just leave the file alone until we delete it.  Changing it further only invites yet more conflicts during upgrade.13:07
gmbjtv: +1; that card’s from before I knew what we needed to do :)13:09
jtvOK13:09
jtvAnd can you confirm that we have no plans to migrate existing bootresources.yaml settings?  It wasn't meant to be very end-user-ish anyway, so I'm not sorry.13:09
jtvThe initialisation code I have up for review gives a user a clean slate, even after upgrading.13:10
* jtv steps out for real13:21
* gmb -> switching locations14:03
jtvallenap, gmb, rvba: I get the impression that "./bin/test.pserv provisioningserver.import_images.tests.test_boot_resources" actually downloads images...14:57
allenapBelt, braces, and belly.14:57
jtvIt goes into the ppc64el boot method, whose install_bootloader starts downloading.14:57
jtvallenap, gmb, rvba: fix is included in https://code.launchpad.net/~jtv/maas/no-bootresources-rewrite/+merge/220046 — who will review it for me?15:11
allenapjtv: +1d15:13
jtvThanks!15:14
jtvgmb: does the plan for ditching bootresources.yaml involve migrating existing config?15:39
gmbjtv: It doesn’t ATM. It definitely *could*, whether it’s an absolute necessity I’m not sure.15:54
gmb(I wonder how many people will have touched their bootresources.yaml)15:54
jtvI see some complications, e.g. "should we also try to migrate from pre-existing images" and "must we go through an API exchange right before importing, just for this upgrade?"15:55
rvbagmb: the API work for boot source is up for review.16:02
gmbrvba: I saw. Thanks :). That’s going to be a nice task to take me to EoD :)16:02
rvbagmb: good luck :)16:02
rvbagmb: I'm sure you can leave some of it for Julian.16:02
gmbNow there’s a thought…16:03
jtvgmb: NodeGroup.bootsource_set.  :)16:06
=== roadmr is now known as roadmr_afk
=== roadmr_afk is now known as roadmr
=== roadmr is now known as roadmr_afk
=== roadmr_afk is now known as roadmr
=== CyberJacob|Away is now known as CyberJacob
=== vladk is now known as vladk|offline
=== roadmr is now known as roadmr_afk
=== roadmr_afk is now known as roadmr
=== roadmr is now known as roadmr_afk
=== vladk|offline is now known as vladk
AskUbuntunot getting impi after installing 14.4 maas, and commisioning servers | http://askubuntu.com/q/46930619:56
=== vladk is now known as vladk|offline
=== vladk|offline is now known as vladk
=== vladk is now known as vladk|offline
=== vladk|offline is now known as vladk
=== vladk is now known as vladk|offline
=== roadmr_afk is now known as roadmr
=== roadmr is now known as roadmr_afk
=== roadmr_afk is now known as roadmr
=== CyberJacob is now known as CyberJacob|Away

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!