jaksi07c8 | hi | 09:20 |
---|---|---|
jaksi07c8 | I'm using cloud-init with the cloud image of ubuntu trusty | 09:22 |
jaksi07c8 | and sometimes it fails to install the packages (packages: section of the cloud-config file) | 09:23 |
jaksi07c8 | it actually fails to configure them, running `dpkg --configure -a` after cloud-init has finished solves the problem | 09:24 |
jaksi07c8 | what could be the cause of this? the environment is the same, and sometimes it works, sometimes it doesn't. | 09:25 |
=== devicenull_ is now known as devicenull | ||
Odd_Bloke | jaksi07c8: Does /var/log/cloud-init.log have anything related? | 09:57 |
Odd_Bloke | jaksi07c8: (And is this an official Ubuntu image, or a derivative?) | 09:57 |
jaksi07c8 | Odd_Bloke: it's the official image | 10:03 |
jaksi07c8 | no, i didn't find anything in the logs | 10:03 |
Odd_Bloke | jaksi07c8: Is it specific packages or can you reproduce with any packages? | 10:04 |
jaksi07c8 | Odd_Bloke: there's a fixed set of packages to be installed | 10:05 |
jaksi07c8 | sometimes all of them are installed | 10:05 |
jaksi07c8 | sometimes none, sometimes just a few | 10:05 |
jaksi07c8 | these are actually our own packages from our own apt repository, so the environment is even more fixed | 10:06 |
Odd_Bloke | jaksi07c8: And these packages definitely reliably install when not using cloud-init? | 10:08 |
jaksi07c8 | is the standard error also present in the logs? | 10:08 |
Odd_Bloke | Perhaps. | 10:08 |
Odd_Bloke | jaksi07c8: Could you paste your cloud-init.log somewhere? | 10:08 |
Odd_Bloke | smoser: https://code.launchpad.net/~daniel-thewatkins/cloud-init/shim_fixes/+merge/269199 fixes some problems that Azure have found with the snappy Azure stuff. | 12:23 |
jaksi07c8 | Odd_Bloke: could it be that cloud-init runs apt-get install in the background, not waiting for it to finish? | 12:40 |
jaksi07c8 | because what I'm experiencing now is that if I shut down the VM right after cloud-init finishes (the login prompt appears), the packages are not installed | 12:40 |
jaksi07c8 | if I wait a little, everything is fine | 12:40 |
Odd_Bloke | jaksi07c8: cloud-init won't block the login prompt appearing. | 12:48 |
jaksi07c8 | mhm... I'm actually replacing an old script which did a preseed-based install of a regular ubuntu server image. The method I used to detect when the installer finishes was that I checked if port 22 was available on the host (sshd has started) | 12:50 |
jaksi07c8 | is there a way to tell if cloud-init has finished everything, and the machine can be safely shut down? | 12:51 |
Odd_Bloke | jaksi07c8: Yeah, that won't work with cloud-init. See http://cloudinit.readthedocs.org/en/latest/topics/examples.html#reboot-poweroff-when-finished | 12:51 |
Odd_Bloke | jaksi07c8: If you need to detect whether cloud-init has finished running while the machine is still up, then you can examine /var/lib/cloud/data/status.json. | 12:54 |
jaksi07c8 | ah, great, thanks a lot for your help! | 12:56 |
Odd_Bloke | :) | 12:57 |
Odd_Bloke | claudiupopa: smoser: Are we going to do the meeting (in 53 minutes; don't panic) in here or in the hangout? | 13:07 |
claudiupopa | I would prefer it to be here, I guess there isn't much to talk for today. | 13:07 |
smoser | Odd_Bloke, here is fine the. | 13:32 |
smoser | then | 13:32 |
* smoser reboots | 13:32 | |
Odd_Bloke | smoser: claudiupopa: Meeting time! | 14:02 |
smoser | woot | 14:02 |
claudiupopa | great. | 14:02 |
smoser | https://review.openstack.org/#/q/project:stackforge/cloud-init+status:open,n,z | 14:03 |
smoser | so the one merge open right now is Cladiu's | 14:03 |
Odd_Bloke | I think that's waiting for another +2 and a workflow +1. | 14:05 |
Odd_Bloke | Unless I have, once again, forgotten how this process works. :; | 14:05 |
smoser | ok. that doesn't have a +1 from Odd_Bloke | 14:05 |
Odd_Bloke | *:p | 14:05 |
smoser | so i was just going to ask that. | 14:05 |
Odd_Bloke | I do remember reviewing this most recent version, not sure why I don't have a review on it. | 14:06 |
Odd_Bloke | Let me log in and see if I have anything I haven't saved. | 14:07 |
smoser | Odd_Bloke, ok. and i'll look at it a bit after this. | 14:07 |
smoser | so lets say that is done, and we should have smoser's comments shortly or a +2/+1 | 14:07 |
smoser | Odd_Bloke, did you do anything for main ? | 14:07 |
claudiupopa | Okay, that sounds good for me. | 14:07 |
* smoser has been on holiday for 12 days (very nice) | 14:07 | |
Odd_Bloke | I was at LinuxCon/Plumbers last week, so I haven't done anything. | 14:07 |
smoser | oh yeah. had forgotten. | 14:08 |
Odd_Bloke | Yep, I'd reviewed everything but hadn't actually +1'd. | 14:08 |
Odd_Bloke | +1'd now. | 14:08 |
smoser | ok then. so if you contineu to look at that, th'ad be good. | 14:09 |
smoser | i have been doing some reporting work in cloud-init 0.7.x and using it in curtin also, much to Odd_Bloke's dismay | 14:09 |
Odd_Bloke | Yep. | 14:09 |
Odd_Bloke | ^_^ | 14:09 |
Odd_Bloke | That yep was to the previous line. :p | 14:09 |
smoser | i will have that submitted to review in 2.0 sometime soon i hope. | 14:10 |
smoser | then we have https://trello.com/b/HoPNdiTI/cloud-init-development-roadmap | 14:10 |
smoser | Odd_Bloke, should take main executable | 14:11 |
smoser | and i'll add a task for 'webhooks reporter' | 14:11 |
smatzek | I was on vacation the week Cladiu's review came out and it got lost in the pile of email. I'll try to get a good review of it this week but don't hold up the merge on my part. | 14:11 |
claudiupopa | I'll probably take the new config module architecture, in order to have soonish a work-in-progress cloudinit v2 that does something. | 14:11 |
smoser | claudiupopa, cool | 14:11 |
Odd_Bloke | cloud-init 2.0 doing something? Controversial. | 14:12 |
Odd_Bloke | I thought we were just going to implement all the fun bits and then do something else. ;) | 14:12 |
smoser | exactly | 14:12 |
smoser | ok. | 14:12 |
smoser | (thanks smatzek for comment above. any help you have is welcomed) | 14:13 |
smoser | thats about all we have then. | 14:13 |
Odd_Bloke | Cool. | 14:13 |
Odd_Bloke | A brief AOB point: do we have a list of everyone we should ping at the start of a meeting? :p | 14:14 |
Odd_Bloke | I can only type sm<TAB> once, so only one of smoser and smatzek are going to get pinged unless we make a list. :p | 14:15 |
smoser | yeah, i know. :) quite rude of smatzek to take collide on my first and last initial. | 14:15 |
smoser | smb is also guilty of that. | 14:16 |
smoser | (/me realizes both of them probably came into existence before him) | 14:16 |
Odd_Bloke | Now you're calling them old; real smooth. :p | 14:17 |
smatzek | claudiupopa: A pre-emptive review on ConfigDrive. If you're going to use 0.7.x as a starting point, can you make util.mount_cb and def find_candidate_devs(probe_optical=True): do the mount and candidate find in a way that can be overridden by the OS distro? | 14:17 |
smatzek | smoser: thanks :) | 14:17 |
smatzek | or have it call to the distro for that support? | 14:18 |
smoser | smatzek, well claudiupopa cares about this legacy OS that comes from some vendor in washington. | 14:18 |
smoser | which apparently doesn't even have a 'mount' | 14:18 |
smoser | so he'll probably abstract it somehow :) | 14:18 |
smatzek | an AIX port of cloud-init exists outside of 0.7.x trunk and this is one of the things that is different in that port. I know if at least 1 other legacy OS, besides the washington one that a port is being created for that also has different ways it handles ISOs. | 14:19 |
smatzek | I hope to put patch sets out to bring the AIX items into 0.7.x and definitely 2.0 once I clear some legal hurdles. | 14:21 |
smoser | cool. | 14:24 |
claudiupopa | Sorry, I was dragged into other meeting for a while. smatzek, smoser: probably it will be abstracted, so we won't care too much about mounting from an interface perspective. | 14:32 |
smoser | claudiupopa, ok... very bad nitpick | 14:33 |
smoser | your commit message width i think is > 80 | 14:33 |
jaksi07c8 | hm.. is there a "right" way to set up static routes with cloud-init? | 16:04 |
jaksi07c8 | I mean I can just run a command like `ip route add 1.2.3.4 via 5.6.7.8`, but that's not the prettiest | 16:05 |
smoser | claudiupopa, commented on your https://review.openstack.org/#/c/209520/ | 16:29 |
smoser | jaksi07c8, no. in the future we expect much better / more aware network knowledge, but for now... i think thats the best you can do . | 16:30 |
smoser | :-*( | 16:30 |
=== jgrimm is now known as jgrimm-away |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!