/srv/irclogs.ubuntu.com/2017/05/01/#juju.txt

vaseyhey folks, i'm trying to bootstrap my juju controller on MAAS, and it's giving me an error bc it can't find a machine that matches its constraints; one of my machines isn't seeing any RAM when  it commissions, is why, but when i try to force a memory constraint to be 0.0 it doesn't work. any ideas?16:30
=== stormmore is now known as Budgie^Smore
rick_hvasey: hmm, so maas has the ram information or maas doesn't have the information?16:48
rick_hvasey: the thing would be to drop the constraint then if it's not going to be able to match. A machine with 0 ram doesn't really mean anything unfortunately. We can file a bug that it should treat it as an empty constraint16:49
rick_hvasey: but just dropping off the ram constraint should be ok?16:49
vaseyrick_h: maas doesn't have the ram information, it thinks it's 0.0GB17:06
vaseywhen i try to add a ram constraint it doesn't override the 3584 MB constraint that's default for the maas bootstrap command17:07
rick_hvasey: can you recommission the node to pick up the memory info?17:09
vaseyi tried, it doesn't help :(17:11
rick_hvasey: hmm, I wonder if we can work around it. Can you add a tag to the machine in MAAS and then use the tags constraint to pick that machine?17:12
rick_hvasey: it should override/skip any memory/default constraints17:13
vaseyoh i'll try that! thanks17:13
rick_hvasey: yea sorry. I'm not sure how to replicate it to test it since MAAS picks up the memory in my setup17:13
rick_hvasey: I wonder why maas is coming up empty17:14
vaseyhttps://pastebin.com/54gTvTz8 here's the output when i try that17:15
vaseylooks like it still wants the memory constraint to be in effect17:15
rick_hbah humbug...hmm17:24
rick_hvasey: can you file a bug here with your commands/etc you've tried out? https://launchpad.net/juju17:24
vaseywill do!17:28
vaseydone17:32
rick_hvasey: ty, sorry I can't think of another way around it atm. So if you say mem= or mem=0.0 it won't work either? can you pastebin the --debug output there?17:32
vaseyboth of those mem constraints don't work unfortunately17:33
vaseyi'll send you the debug in a sec17:33
vaseyrick_h: https://pastebin.com/YSGrp4tZ17:37
rick_hvasey: hmm, so it seems it just doesn't listen to them at all then17:37
rick_hvasey: k, sorry for the trouble. I'd love to get the info into maas and while we could work around this after bootstrap with direct placement, we can't do that on bootstrap. :/17:38
rick_hvasey: I guess you could try the the manual provider route but that's obviously not the smooth UX we're shooting for17:38
vaseyrick_h: how do you do the manual provider route?18:01
rick_hvasey: so you'd need to turn on the machines through maas and bootstrap to it over ssh. https://jujucharms.com/docs/stable/clouds-manual walks you through it18:02
vaseythanks18:02
Budgie^Smoreo/ juju world19:11
kwmonroeverterok: how/where did you deploy spark when it resulted in http://paste.ubuntu.com/24494765/?21:32
verterokkwmonroe: hi21:42
verterokkwmonroe: in canonical DC21:43
verterokkwmonroe: using a mojo spec, but nothing fancy, just spark + nrpe21:44
verterokkwmonroe: let me try with a clean mojo workspace21:46
kwmonroehmm, it's weird verterok, spark-34 tested clean last week across the clouds (and hasn't changed since):  http://bigtop.charm.qa/cwr_bundle_spark_processing/63/report.html21:47
verterokkwmonroe: maybe it's failing because of the restrictions we have in our DC?21:49
verterokkwmonroe: gimme a couple, the deploy is running21:49
verterokkwmonroe: same error: spark/0*  error     idle   0        10.50.84.52            hook failed: "install"21:51
kwmonroeverterok: yeah, i'm going to guess it's a restricted network thing.. and a good enough reason to hold the un-prom for apache-* until we can fix the current stuff.21:54

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