=== pquerna_ is now known as pquerna === evilissimo1 is now known as evilissimo [13:21] meh, still no openstack community newsletter === hatchetation_ is now known as hatchetation [20:14] hey guys, im using 0.7.4 and noticed that with the config drive source that when it checks the uuid for previous installs it fails because the length of the uuids differ (extra white space) so it runs on_first_boot everytime, is this a known bug (maybe fixed already) [20:14] ? [20:27] example http://pastie.org/8769066 === harlowja is now known as harlowja_away [20:36] apologies if I missed a responce, the web client keeps disconnecting me === harlowja_away is now known as harlowja [21:08] smathews, why does it find different content in the uuid ? [21:09] smathews, ie, that doesn't look like insane behavior from cloud-init's perspective. why should it be expected to trim data ? [21:09] my guess is that when it writes there is an extra white space at the end of the uuid in the file [21:12] I dont think the logic here is incorrect, or that it should have to trim [21:12] maybe the write of the uuid is incorrect [21:14] also I think this is related, since if the uuids were matching it wouldnt run the network config https://bugs.launchpad.net/cloud-init/+bug/1275098 [21:15] on subsequent launches that is [21:29] util.write_file(iid_fn, "%s\n" % iid) [21:29] looks like its from the newline [21:29] thoughts? [22:20] smathews, it looks like you're right. [22:20] harlowja, ^ [22:21] hmmm, newlnes! [22:22] heh [22:25] https://code.launchpad.net/~harlowja/cloud-init/strip-iid-newline/+merge/208028 [22:25] so, while we are on the topic of config drives, can I make a request for in on_first_boot to write the files before applying the network config? [22:25] this would be helpful for when writing static routes [22:26] seems fair [22:26] any objections smathews ? [22:26] smoser ^ [22:26] lol [22:26] heh [22:26] i assume smathews doesn't have objects :-P [22:27] hmmm I generally dont object with myself... [22:27] although lol [22:28] :) [22:28] is it not easy to not write the carraige return? [22:28] clearly strip() should be safe [22:28] but we shoud just not write baad data [22:28] or even bad data [22:28] harlowja, one othe rhting i meant to mention to you. [22:28] smoser likely it already exists out there with "\n" for quite a few versions, so better to just strip? [22:28] the datasource for openstack right now reads from "latest" [22:28] which is imo wrong [22:28] ? [22:29] i'm pretty sure it tries "latest" in the metadata service. [22:29] this is somewhat memeory, but there were 2 things i wanted to fix there. [22:29] k [22:29] ya, which version to use? [22:29] a.) i thought it was reading from "latest" by default. as opposed to GRIZZLY or HAVANA (their YYYY-MM-DD respectively) [22:29] sure, seems so [22:30] b.) it was warning when it tried one. warn goes to stdout, which goes to console and log as a warn [22:30] and that wasn't really a failure case. [22:30] agreed [22:30] repairing [22:32] smoser, thoughts on writing files before apply network in on_first_boot? [22:38] smathews, i dont follow [22:38] writing files as in 'write_files' ? [22:39] yeah, in on_first_boot apply_network happens before write_files [22:39] Id like to switch that [22:40] it'd be handy for writing static routes and such [22:42] smathews, i have to thikn about it. sorry . have to run right now [22:42] no worries [22:43] thanks for your help! [22:43] smoser https://code.launchpad.net/~harlowja/cloud-init/ds-os-adjust/+merge/208031 [22:43] whenver u get free === ctracey` is now known as ctracey|away