=== harlowja_ is now known as harlowja_away | ||
HanSolo | Hello guys | 10:47 |
---|---|---|
HanSolo | what is the best way to add my elastic ip in /etc/hosts with cloud-init please | 10:47 |
HanSolo | ? | 10:58 |
smoser | HanSolo, theres nothing that can do thta for you. :-( | 13:05 |
Odd_Bloke | smoser: Thanks for the merge yesterday. :) | 13:42 |
Odd_Bloke | smoser: Here's another little one: https://code.launchpad.net/~daniel-thewatkins/cloud-init/lp-1311463/+merge/253362 :) | 13:43 |
smoser | Odd_Bloke, ok. thank yo ufor writing tests. | 13:43 |
Odd_Bloke | :) | 13:45 |
HanSolo | smoser: I have to use this (ugly) kind of workaround then ? http://ternarylabs.com/2010/09/15/automatically-configure-hostname-for-new-ec2-instances/ | 14:38 |
HanSolo | I know this is beyond scope of cloud-init sorry for asking :/ | 14:39 |
smoser | HanSolo, its not really beyond the scope. | 14:39 |
smoser | i'd like to have a solution for you | 14:39 |
HanSolo | thank you smoser, this is still a great product | 14:41 |
=== harlowja_away is now known as harlowja_ | ||
harmw | harlowja_: ever seen this with kvm? https://bugzilla.redhat.com/show_bug.cgi?id=1202990 | 17:59 |
harlowja_ | harmw don't think i've seen that | 18:00 |
harlowja_ | but we don't run on fedora, so thats maybe why | 18:00 |
harmw | ah, well, CentOS 7 showed the same | 18:00 |
harlowja_ | ah | 18:01 |
harlowja_ | haven't seen that | 18:01 |
harmw | are you aware of certain ovs specifics one should be setting? | 18:01 |
harlowja_ | not off the top of my head | 18:01 |
harmw | i'm using an mtu of 1400 here btw, for what its worth | 18:01 |
harlowja_ | harmw if u jump on #openstack-operators u may find some folks there that have some idea | 18:14 |
harmw | oh sure, thanks | 18:14 |
harlowja_ | np | 18:14 |
harlowja_ | someone there might of seen this | 18:14 |
harmw | I'm gonna try if I can get some results from -not- running through nova | 18:14 |
harlowja_ | k | 18:14 |
harmw | eg, dropping a lot of the extra kvm parameters | 18:15 |
harlowja_ | ya, maybe u'll figure it out | 18:15 |
harmw | probably not though :P | 18:30 |
harlowja_ | lol | 18:36 |
stupidnic | Could somebody help me troubleshoot a long period on cloud-init modules:config | 19:26 |
stupidnic | I run cloud-init on a centos6 image and it get's an IP address quickly but then it just sits there for a long duration before completing | 19:29 |
stupidnic | http://pastebin.com/XkwiUxsN | 19:29 |
stupidnic | that's what the logs show | 19:29 |
=== shardy is now known as shardy_z | ||
smoser | stupidnic, i suspect /var/log/cloud-init.log has some WARN in it. | 23:26 |
smoser | can you paste that ? | 23:26 |
stupidnic | smoser: it does not | 23:27 |
smoser | hm.. | 23:27 |
stupidnic | let me jump into that server again | 23:27 |
smoser | its poling something. | 23:27 |
stupidnic | that was my thought too | 23:27 |
stupidnic | seemed like a timeout or something | 23:27 |
stupidnic | cloud-init.log is zero length | 23:28 |
stupidnic | I can show the entire cloud-init-output.log if you like | 23:28 |
stupidnic | http://pastebin.com/gimfKReE | 23:30 |
stupidnic | nothing much really | 23:30 |
stupidnic | there is just that huge 500+ second delay between the ci-info output and the modules:config run | 23:30 |
stupidnic | I can clone the base image and turn on more debugging, I just don't know how to do that actually | 23:31 |
smoser | stupidnic, the log is getting swallowed somewhere. | 23:35 |
smoser | oh.. syslog. | 23:35 |
smoser | try looking in /var/log/syslog or wherever syslog messages would go. | 23:35 |
smoser | grep -r "cloud" /var/log | 23:35 |
stupidnic | coming up empty | 23:36 |
stupidnic | well not empty empty | 23:36 |
stupidnic | but nothing new | 23:36 |
smoser | odd. well, anything important here ? | 23:36 |
smoser | ie, do you care particularly about the instance | 23:37 |
smoser | 2 things to try | 23:37 |
smoser | a.) edit /etc/cloud/cloud.cfg.d/05_logging* | 23:37 |
stupidnic | It's a customer instance | 23:37 |
smoser | then yes, yes you do :) | 23:37 |
smoser | reproduce somewhere else ? | 23:37 |
stupidnic | but I can dump the base image out of glance | 23:37 |
stupidnic | and then fiddle with that | 23:37 |
stupidnic | and the put it back into glance and use it as a base for testing instances | 23:38 |
smoser | oh. can you see the console log of the instance ? | 23:38 |
smoser | nova console-log ... | 23:38 |
stupidnic | Yeah it's more of the same sadly | 23:38 |
smoser | that may have WARN on it too | 23:38 |
stupidnic | the original pastebin was taking from that | 23:38 |
smoser | well, debug messages woudl be useful. | 23:39 |
smoser | so that 05_logging.cfg | 23:40 |
smoser | just comment otu: | 23:40 |
smoser | - [ *log_base, *log_syslog ] | 23:40 |
smoser | that line. | 23:40 |
smoser | and then it should log directly to the file | 23:40 |
stupidnic | k | 23:40 |
stupidnic | let me dump the image and get that taken care of | 23:40 |
stupidnic | smoser: okay... what should I be commenting out here? just the stuff under log_cfgs: section? | 23:45 |
smoser | just that line. | 23:46 |
stupidnic | alright, done... let me load the image back into glance and spin up an instance | 23:47 |
stupidnic | well of course... it didn't do it this time | 23:52 |
stupidnic | cloud-init took 22 seconds from start to finish | 23:52 |
smoser | do you see log too ? | 23:52 |
stupidnic | yes | 23:53 |
stupidnic | the cloud-init.log is over 140KB in size | 23:53 |
stupidnic | About the only difference this time was the size of the instance and the tenant | 23:54 |
stupidnic | client instance was 1TB this was only 10GB | 23:54 |
stupidnic | possible that the size difference was the delay? | 23:55 |
smoser | oh. well, was it the first boot ? | 23:55 |
stupidnic | Yes | 23:55 |
smoser | yeah. cloud-init resizes the disk | 23:56 |
smoser | and on an older kernel, that is very slow. | 23:56 |
stupidnic | I thought that occurred in the init ramdisk | 23:56 |
smoser | on older kernels, it will rewrite the partition table in the ramdisk | 23:56 |
stupidnic | at least in Centos 6 | 23:56 |
stupidnic | right | 23:56 |
smoser | (on newer ones it does that in cloud-init too) | 23:57 |
stupidnic | yeah this is a centos 6 image | 23:57 |
stupidnic | so 2.6.32 kernel | 23:57 |
smoser | < 3.2 kernel is stupid slow resizing. | 23:57 |
smoser | i tihnk its actually 3.10 that is , not kidding, 1000x faster. | 23:57 |
stupidnic | hmmm... might be worth going with the centos lt kernels then | 23:58 |
stupidnic | Something to consider. I don't want to fragment my images too much though | 23:58 |
smoser | is not really taht simple (its hiding some of the operations i think) but, a 'resize2fs from 2G -> 1TB' will be probalby 1000x faster on nwere kernel. | 23:58 |
stupidnic | last thing I need is to support five variants of each OS | 23:58 |
smoser | yeah. i'd probalby live with it. | 23:58 |
stupidnic | Let me try a larger instance and see what it does | 23:59 |
stupidnic | I have SAN to burn ;) | 23:59 |
smoser | https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1179610 | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!