[19:44] powersj: did you test console output ? [19:45] in your ec2 [19:45] I added the call to grab it, however as you will note about my comment it isn't always ready to get pulled [19:45] seems like it should return a base64 encoded string [19:45] http://boto3.readthedocs.io/en/latest/reference/services/ec2.html#EC2.Instance.console_output [19:45] the latest commit got it from: https://git.launchpad.net/~powersj/cloud-init/commit/?id=677c60738d026a9bc1272c3b6e2366ce4bdc98f8 [19:45] ie, if you *did* get something i think it will be base64 [19:46] also the base64 encoding doesn't apply since "If using a command line tool, the tool decodes the output for you." [19:46] when using the aws-cli == command line tool [19:47] right [19:47] you're *not* using it [19:47] so you need to decode [19:47] or using the python library seems to cover that as well [19:47] that seems wrong [19:47] same for user-data... we don't encode it [19:47] https://eucalyptus.atlassian.net/browse/TOOLS-407 [19:48] err.. the redux [19:48] https://eucalyptus.atlassian.net/browse/TOOLS-675?attachmentViewMode=gallery [19:49] basically, though, i'd expect that the content that went to the console of the device is not necessarily anything that be represented losslessly as a string [19:49] it is bytes [19:49] so if ec2 did an encode to utf8, that'd be lossy [19:51] smoser: your comments stem from my use of .encode() on the output? [19:52] have you tested this? [19:52] yes ran it through all the tests, only one fails and we need to fix that test [19:52] let me try again in case I just never got any output at all [19:52] because i think what you will be '.encode()'ing is b'aGVsbG8gd29ybGQK' [19:53] so sure, it will work, buit the console log will be base64 encoded string [19:53] er,...bytes. [19:54] i have to run, i'll take a further look later. [19:59] smoser: it produces a string [19:59] https://paste.ubuntu.com/26321229/ [20:00] here is also an example of no output: https://paste.ubuntu.com/26321219/ [20:03] here is part of the raw string: https://paste.ubuntu.com/26321243/ [20:07] powersj minor comments on my last pass of your ec2 branch posted [20:07] approved on my end once those nits are reviewed and handled or rejected by you [20:09] blackboxsw: thank you, I'm +1 on all of those [20:12] hey smoser, I'm working on using url_helper instead of raw requests, but I need infinite retry support so I was planning on using max_wait==None. I wanted to make sure that is okay, as well as wanted to point out that right now, using the default parameter of None will trigger a somewhat unexpected exception [20:13] def timeup(max_wait, start_time): return ((max_wait <= 0 or max_wait is None) or (time.time() - start_time > max_wait)) [20:13] the problem is the short circuit check for a None parameter should happen before the none to int comparison. (this is in url_helper) [20:15] smoser: in addition to the above, take a look at: https://paste.ubuntu.com/26321285/ [20:39] blackboxsw: changes pushed [20:39] spelling fix, docstring format, try/except change, and more details in debug messages [20:42] powersj: looks good, just saw another highlighted in your latest changest. 'deleting secuirty' in tests/cloud_tests/platforms/ec2/platform.py [20:42] all good otherwise. [20:44] blackboxsw: thx are you using a spell checker of some sort? [20:45] I have two of them, and no prescription glasses ;) [20:45] lol [20:45] mine don't work very well [20:45] nope just reading through things generally. that's why I missed the typos first time around (I should use a spell checker) ispell or something [21:01] blackboxsw: https://paste.ubuntu.com/26321529/ may have some test escapes again [21:01] that was on my local branch, let me see if the nightly run shows the same [21:02] actually ignore me... I am looking at the decimal wrong doh [21:02] coffee time [21:02] powersj: hrm, yeah that is taking a bit long..... I thought when we saw bad escapes we'd see > 1 second (but that was probably with retries etc) [21:02] yeah I saw that and went 5 seconds?! [21:02] but it isn't 5 seconds :P [21:04] powersj: yeah I went on a date to zquila over break and had a decimal place issue. we almost ordered 2 margs that were $75 each. thank goodness the waitress asked us to double check [21:04] hahaha [21:05] I can't believe enough people would actually order a $75 marg to warrant putting something like that on the menu [21:06] #wrongchannel [21:06] :) [21:09] LOL [21:58] powersj: then the bug is in boto3 :-( [21:58] bug? [21:58] i'll file it upstream. [21:59] you believe it should return bytes over a string? [21:59] it is taking a base64 encoded blob and decoding it into a string [21:59] and losing data along the way [22:01] https://github.com/boto/botocore/blob/develop/botocore/handlers.py#L142 [22:01] yeah, thats the bug :) [22:08] https://github.com/boto/botocore/issues/1351 [22:08] thanks for digging, powersj [22:09] smoser: heh thanks for calling it out :) [22:09] I am always amazed by your ability to pull out pugs from years ago heh :P [22:09] firefox history [22:10] i think euca must not use boto3 [22:10] otherwise it would never have seen the bug that i reported (twice) [22:10] i know it didn't origintally use boto3, but i figured it might have moved by now. but the project is kind of dead now. [22:13] dojordan_: yes, that seems sane. 0 might also make sense. [22:14] powersj: fwiw i'm not just making this stuff up. the data that systemd writes to /dev/console is almost all the time *not* utf-8 during a boot cycle on ubuntu at least. [22:14] i only know this because of tools that assumed it was not working :) [22:15] ok good to know, my quick look at the data made me go looks like normal output and move on [22:27] powersj: i hit 'submit' on review [22:27] i think i might be missing something though... i just went through my last round of comments, and it didn't look like you addressed many of them. [22:29] smoser: hmm thought I did [22:29] I'll look [22:40] smoser: looks like I misunderstood your comment about snapshots, otherwise hit all your previous requests [22:40] I'll get these fixes done === simpoir is now known as simpoir|afk === simpoir|afk is now known as simpoir [23:40] smoser: pushed updates, with a question on the review