[03:31] <bigjools> roaksoax: can you bless this please? https://code.launchpad.net/~julian-edwards/maas/packaging/+merge/211668
[08:22] <rvba> bigjools: is there something fundamentally wrong with https://code.launchpad.net/~rvb/maas/bootimage-ui/+merge/211761 or did you just forget to mark it 'Approved'?
[08:23] <jtv> Where did Oleg add his comments?  I don't see them in the new-import-script-integration branch.
[08:23] <jtv> Oh, wait
[08:23] <rvba> jtv: https://code.launchpad.net/~strikov/maas/maas-new-metadata-format/+merge/210843
[08:23] <jtv> trunk now, right?
[08:24] <jtv> Thanks.
[08:24] <jtv> Do we have parallel branches now?
[08:24] <jtv> I've been working off the new-*-integration one.
[08:25] <rvba> jtv: yeah, we need to integrate Oleg
[08:25] <rvba> 's work
[08:26] <jtv> Should we start by just making all of it live in trunk, even if not active yet?
[08:27] <jtv> Keep the old and the new entries in pserv.yaml, etc.
[08:27] <jtv> Or will it be too hard?
[08:43] <rvba> jtv: I think it's doable.
[08:44] <rvba> And probably the best option to avoid the big bang approach.
[08:44] <jtv> Yeah.  It'll still be one functionally, but at least we won't be messing around with conflicting branches.
[08:44] <jtv> I went through a weird jet-lag phase earlier today, and I was just constantly wondering what was where.  Scary!
[08:45] <jtv> Trunk is nice and safe.  :)
[08:53] <bigjools> rvba: I forgot, sorry.
[08:53] <rvba> no worries
[08:54] <bigjools> rvba, jtv: I'd like to get oleg using the integration branch
[08:54] <bigjools> however I emailed him earlier and said we'd just take over the work
[08:54] <jtv> Then we should figure out how to get his latest work into either maas or the integration branch, and the integration branch into trunk.
[08:56] <bigjools> unless we changed any of the same code that is in trunk, we can just get rid of the one in trunk
[08:56] <bigjools> did anything change in the integration branch?
[08:57] <jtv> Not yesterday or today.
[08:57] <jtv> I've been pulling it and not seeing changes.
[08:57] <jtv> Nothing since Tue 2014-03-18 10:53:34 +0100
[08:58] <bigjools> ok
[09:06] <jtv> rvba: I'm trying to run the import script from the integration branch, against source http://maas.ubuntu.com/images/ephemeral-v2/daily/ — but it just fails because the snapshot directory is never created.  Do you get the same?
[09:06] <rvba> jtv: I didn't try using maas.ubuntu.com yet.
[09:06] <bigjools> I am about to try
[09:06] <rvba> I've always used the fake data thus far.
[09:06] <rvba> jtv: I'll try now.
[09:07] <jtv> FWIW I got the same with the test repos, so I expect it's something I'm doing wrong.
[09:26] <rvba> jtv: bigjools: I've integrated Oleg's most recent fixes to the import script in our integration branch.
[09:26] <bigjools> awesome
[09:26] <jtv> Great, thanks.  I'll re-merge.
[09:26] <bigjools> let's get him using it as well
[09:27] <bigjools> meanwhile, my "make install-dependencies" is still going 20 minutes later on this lcy02 instance
[09:27] <rvba> Yeah, canonistack is a mess these days.
[09:27] <rvba> Machines on lcy02 and lcy01 are slow as hell.
[09:28] <bigjools> popular I guess
[09:28] <rvba> (All the bootimages stuff is landed, I need to QA it but it should be quick)
[09:31] <bigjools> maas seems to have about 6 million dependencies
[09:34] <rvba> jtv: when I merge trunk into the integration branch I get a conflict related to the 'label' work.. could you have a look?  You can probably fix this quicker than me.
[09:35] <jtv> Sure
[09:35] <rvba> Ta
[09:35] <jtv> I don't know about the armhf template though.
[09:37] <rvba> We can take Oleg's version, 'highbank' is now explicitly supported by the 'generic' kernel.
[09:37] <rvba>    "version": "14.04",
[09:37] <bigjools> what/where do you configure the new source in the pserv.yaml?
[09:37] <rvba>    "subarches": "generic,highbank,hwe-p,hwe-q,hwe-r,hwe-s,hwe-t",
[09:37] <rvba>    "release": "trusty",
[09:37] <rvba>    "arch": "armhf"
[09:38] <rvba> bigjools: not sure what you mean…?
[09:38] <jtv> Comma-separated string?  Not a list?
[09:38] <bigjools> rvba: the new simplestreams source, where do I configure it in oleg's changes?
[09:38] <jtv> etc/maas/pserv.yaml
[09:38] <rvba> bigjools: http://paste.ubuntu.com/7124389/
[09:38] <rvba> See the top of the diff in https://code.launchpad.net/~maas-maintainers/maas/new-import-script-integration/+merge/211470
[09:39] <bigjools> guys, I can see that much, but where in that config!
[09:39] <rvba> The path.
[09:39] <rvba> Sorry, I don't think I understand your question.
[09:39] <jtv> Going to be hard to resolve that conflict when tests don't pass in the existing branch...
[09:39] <rvba> jtv: yeah, I know :/
[09:40] <jtv> If tests don't pass anyway, just go with the trunk version...
[09:43] <jtv> The armhf template conflict is just two identical changes conflicting.  I thought they'd made bzr accept that without complaining.
[09:45] <bigjools> formatting difference?
[09:45] <bigjools> trailing spaces?
[09:46] <jtv> No.
[09:46] <jtv> I ran a diff.
[09:46] <bigjools> :(
[09:47] <bigjools> well I tried to run the new script and got this
[09:47] <bigjools> IOError: [Errno 2] No such file or directory: u'/var/lib/maas/boot-resources//snapshot-20032014-094738//maas.meta'
[09:49] <jtv> Same for me.
[09:49] <jtv> That's the snapshot directory I mentioned earlier.
[09:50] <jtv> It looks as if simplestreams is supposed to create it implicitly while synchronising.
[09:50] <jtv> By the way, note the 20032014... yes that _is_ a date but it's little-endian.
[09:53] <rvba> Hi strikov!
[09:54] <strikov> rvba: Hey :)
[10:19] <rvba> strikov: that's the integration branch: lp:~maas-maintainers/maas/new-import-script-integration
[10:28] <gmb> bigjools: To confirm: when you had your empty drop down problem, what fixed it? Re-running import-(eph|pxe)?
[10:30] <bigjools> gmb: in my case it was putting the tftp images back in the right place
[10:30] <gmb> bigjools: But wouldn't running the script again had the same effecT?
[10:30] <gmb> (I'm asking because the error should give some instruction as to how to get out of the problem state)
[10:30] <bigjools> gmb: yes, but not in my case as I had a screwy package
[10:30] <gmb> Well, we don't care about that. :)
[10:30] <gmb> Okay, cool.
[10:30] <bigjools> yes, the instruction should be the same as the bug warning banner about missing boot images
[10:31] <bigjools> big*
[10:31] <gmb> Right. That's what I'm going to steal :)
[10:32] <bigjools> gmb: parfait
[10:34] <allenap> Grr.
[10:34] <rvba> allenap: \o/
[10:34] <rvba> ;)
[10:34] <bigjools> wb allenap
[10:34] <allenap> Don’t talk to me.
[10:34] <allenap> ;)
[10:35] <bigjools> so gmb, allenap, rvba, jtv: I shall make sure kanban cards are complete for the list on that doc
[10:35] <bigjools> and then please feel free to start taking them
[10:37] <bigjools> allenap: here's a good reason to start with a new config file, we need to write into it some config based on the current bootimages
[10:37] <bigjools> does the yaml writer maintain comments?
[10:37] <allenap> bigjools: Nope, it doesn’t.
[10:37] <bigjools> uh huh
[10:37] <allenap> Afaik.
[10:37] <bigjools> :)
[10:39] <allenap> bigjools: However, you can just append to the yaml file and it’ll dtrt.
[10:39] <bigjools> allenap: true!
[10:40] <allenap> When it comes to in-place updates to config files, there’s nothing really compelling out there. Configobj is the closest I’ve seen, but that’s kind of baroque.
[10:40] <jtv> Do we need to store this in a config file at all?
[10:40]  * allenap can’t remember exactly what we’re storing.
[10:41] <jtv> Parameters for the imports: arches, releases, source URLs.
[10:41] <jtv> Now, the source URLs we need to have somewhere.
[10:41] <lifeless> configObj is what bzr used
[10:41] <jtv> But the rest I guess could be generated from a form.
[10:41] <rvba> allenap: http://paste.ubuntu.com/7124600/
[10:41] <jtv> Hi lifeless
[10:42] <jtv> What I had in mind originally was: import link goes to a form page, where you can enter the releases and architectures you want, and _that_ gets pre-populated based on existing boot images.
[10:42] <jtv> Oh, that doesn't work with multiple simplestreams sources does it?
[10:43] <rvba> This is all the test failures we've got on the integration branch http://paste.ubuntu.com/7124602/
[10:43] <jtv> I would suggest people start fixing in small branches...
[10:43] <bigjools> wotcha lifeless
[10:44] <rvba> bigjools: can you land your packaging branch?  I can't QA my changes right now because of the new dependency.
[10:45] <bigjools> rvba: yes!
[10:45] <rvba> jtv: yes, the failures can be easily divided into "topics"
[10:45] <rvba> bigjools: ta
[10:45] <jtv> rvba: so you never had that error where the snapshot directory did not exist?
[10:45] <rvba> jtv: no
[10:45] <jtv> And you used that config you pasted to Julian earlier?
[10:47] <rvba> jtv: no, I was using the test data so that path was different.
[10:47] <jtv> What path?
[10:47] <jtv> Oh, the URL?\
[10:48] <jtv> What did you use?  The path to the repo1 (or repo2) directory?
[10:48] <rvba> Yeah, the URL.  Named 'path' in the config.
[10:48] <jtv> file:// URL?
[10:48] <allenap> lifeless: Ah, interesting, that makes me want to look again.
[10:48] <rvba> Yes.
[10:48] <jtv> That didn't work for me either.  Maybe it was the various problems in pserv.yaml...
[10:49] <jtv> Because I noticed that you passed a comma-separated list for subarches, not a list like in the integration branch.
[10:50] <jtv> Ugh.  When I try it against the simplestreams-v2 daily URL, it breaks because it looks for an sjson index.
[10:50] <jtv> Which isn't there.
[10:51] <bigjools> rvba: I can help fix tests
[10:51] <bigjools> let me just get the int  branch
[10:51] <jtv> strikov: do you know if, for manual testing, we can make the import script use index.json (as opposed to index.sjson) and skip any attempt at verifying signatures?
[10:52] <rvba> bigjools: cool, the card contains a link with all the test failures.
[10:52] <rvba> bigjools: I'm fixing the kernel opts-related failures now.
[10:52] <strikov> jtv: its better to put signatures in place, i think
[10:52] <strikov> jtv: let me tell how in a moment
[10:52] <strikov> *tell you how
[10:52] <bigjools> ta
[10:52] <jtv> strikov: thanks.  It's better with signatures of course, but right now, there aren't any and we do need try out some manual runs.
[10:52] <allenap> lifeless: While you’re there, would you be interested in something like https://code.launchpad.net/~allenap/maas/run-isolated-with-debugging/+merge/211803 in upstream subunit?
[10:53] <rvba> https://code.launchpad.net/~rvb/maas/integ-fix-tests/+merge/211903
[10:53] <strikov> jtv: bzr branch lp:simplestreams simplestreams
[10:53] <strikov> cd simplestreams
[10:53] <strikov> make gnupg
[10:53] <strikov> ./tools/tenv js2signed /home/ubuntu/boot-resources-v2/
[10:54] <strikov> jtv: that will sign your metadata with simplestream's default key which is in ./gpg/ folder
[10:54] <jtv> strikov: not what I needed!
[10:54] <strikov> jtv: then you can add this key to your keyring and all the checks will pass
[10:54] <jtv> But I'm trying to import a stream that doesn't have any signatures at all.
[10:55] <strikov> jtv: js2signed will add all the signatures automatically (create sjson from json)
[10:55] <jtv> Oh, it inserts that on my end somehow?
[10:55] <jtv> Instead of on the stream I'm trying to import from?
[10:55] <strikov> jtv: which meta do you use? self-generated?
[10:56] <jtv> No, Scott's online URL.
[10:56] <jtv> http://maas.ubuntu.com/images/ephemeral-v2/daily/streams/v1/
[10:56] <strikov> jtv: Ah. Try mine: http://162.213.35.97/boot-resources-v2/
[10:56] <strikov> jtv: it's basically the same but with signatures
[10:57] <strikov> jtv: you need to integrate simplestreams key though
[10:57] <jtv> Thanks.  That will give me an exact mirror of Scott's otherwise?
[10:57] <bigjools> rvba: can you talk me through that MP
[10:57] <strikov> jtv: let me see if scott made any changes yesterday
[10:57] <jtv> Thanks.
[10:57] <jtv> AIUI labels were added.
[10:58] <rvba> bigjools: this is to account for this http://paste.ubuntu.com/7124669/
[10:58] <bigjools> rvba: I guess that's pretty simplified!
[10:58] <rvba> Yeah, we don't need to patch things up anymore.
[10:58] <strikov> jtv: well, he did (and that's good), we indeed need to disable sign check then
[10:59] <bigjools> rvba: but this means we only keep the most recent locally?
[10:59] <rvba> Yes.  That's what the import script does.
[10:59] <jtv> strikov: yeah... for this kind of thing I really want to be close to the source, so we don't get everything jusssst right — for an obsolete version.
[10:59] <rvba> Also, the 'info' file is now completely gone.
[11:00] <jtv> rvba: get_ephemeral_name also gets a label, of course, in the branch I'm working on.
[11:00] <rvba> Right.
[11:00] <strikov> jtv: to disable sign check we need to (1) change index.sjson to index.json inside script and (2) implement stub policy handler for simplestreams
[11:00] <strikov> jtv: does it work for you?
[11:01] <jtv> strikov: sure, it's only for a temporary hack.  Thanks!
[11:01] <strikov> jtv: give me a few moments
[11:01]  * jtv gives strikov a dozen moments
[11:08] <bigjools> rvba: fwiw not sure I'd bother with MPs for the int branch fixes
[11:08] <rvba> bigjools: wfm
[11:08] <bigjools> rvba: but for safety, did you set append_revisions_only?
[11:08] <rvba> bigjools: no
[11:08] <bigjools> rvba: better do it :)
[11:08] <bigjools> since we're all going to be pushing up
[11:09] <rvba> bigjools: how do I do that?
[11:11] <rvba> bigjools: found it; done.
[11:11] <bigjools> rvba: you were quicker than me
[11:11] <bigjools> sftp to the branch location basically
[11:11] <rvba> bzr config -d bzr+ssh://bazaar.launchpad.net/~maas-maintainers/maas/new-import-script-integration/ append_revisions_only=True
[11:12] <bigjools> http://stackoverflow.com/questions/5413602/monotonically-increasing-bazaar-trunk-revision-numbers
[11:15] <lifeless> allenap: hmm
[11:15] <lifeless> allenap: so with v2 you can debug through the stream
[11:15] <allenap> lifeless: Ah ha, okay. We ought to upgrade at some point then.
[11:16] <lifeless> allenap: but, that wouldn't stop the dup madness
[11:16] <lifeless> allenap: so I think that this is a simple bug and we should fix it in subunit
[11:16] <lifeless> separately to any discussion about v2 upgrades
[11:17] <allenap> lifeless: Okay, I’ll put a patch together.
[11:17] <lifeless> let me look a little more
[11:17] <lifeless> ok
[11:17] <lifeless> so we replace fd 1
[11:17] <lifeless> which means sys.stdout will still go to that fd
[11:18] <lifeless> and that fd does need to be the fd of the parent pipe, or you won't get the test data
[11:18] <lifeless> so yeah - I think this is a case for upgrade the connection to v2 and route stdin and out to the child
[11:18] <lifeless> via v2 packets
[11:18] <lifeless> should be fairly straight forward
[11:18] <lifeless> get you debugging
[11:19] <allenap> I think I need to read up on v2’s design.
[11:23] <strikov> jtv: https://pastebin.canonical.com/106840/
[11:23] <strikov> jtv: it disables sign check and switches from md5 to sha256
[11:23] <jtv> Thanks!  I'll feel a lot better knowing that I can try things out against Scott's latest "real" data.
[11:24] <jtv> allenap, bigjools, gmb, rvba: see strikov's paste above — the ephemeral-vs stream is not signed yet, so Oleg's patch should help us work around that.  Do Not Commit.  :)
[11:25] <strikov> jtv: Right. Give me one more moment please -- it crashes with Scott's data due to some reason (not related to sign check)
[11:25] <jtv> Whoopsie
[11:25] <bigjools> yay? :)
[11:26] <jtv> It's one step further than what I got.  :)
[11:26] <rvba> jtv: we should probably steal the signature-related code from the old ephemeral import script.
[11:27] <strikov> jtv: heh, that's not my problem though -- so of the entries in Scott's meta doesn't contain checksums
[11:27] <rvba> And maybe extend it a bit to allow unchecked downloads.
[11:27] <strikov> *some of
[11:27] <jtv> rvba: For Freudian reasons, you probably know more about that code than I do...
[11:28] <jtv> rvba: I don't think we have much signing-related code in there anyway.
[11:28] <rvba> jtv: just an option to choose the keyring.
[11:28] <bigjools> rvba: I am thinking that we can get rid of provisioningserver.import_images.tests.test_ephemerals_script entirely
[11:29] <rvba> bigjools: not entirely, there are tests for helpers we still use in there.
[11:29] <rvba> uec2roottar thingy, compose_filters
[11:30] <bigjools> rvba: ok I'll do it
[11:30] <strikov> jtv: well, i'd say that Scott's meta is broken in many ways
[11:30] <jtv> ?
[11:31] <rvba> Yay
[11:31] <strikov> jtv: not all entries contain checksums (which is required for my script because its cache is checksum-indexed)
[11:31] <jtv> Ouch
[11:32] <strikov> jtv: that's definitely a requirement for meta to provide these checksums -- so Scott just didn't fix it yet
[11:32] <rvba> strikov: it's important to keeps a list of all these problems with the new metadata.  Would you mind adding it to the document we looked at during the call?
[11:32] <strikov> rvba: well, i don't think that's a real problem because Scott didn't tell me that his meta is ready yet
[11:32] <strikov> rvba: so, he's working on it right now
[11:34] <rvba> strikov: I know, but we should keep track of what creates problems for the import script.  So that we can make sure things are fixed when Scott releases a new version.
[11:34] <strikov> jtv, rvba: how about not pulling resources w/o checksums available?
[11:34] <strikov> jtv, rvba: that's a hack but it allows us to move forward
[11:34] <jtv> You mean just ignoring them?  Sounds reasonable to me.
[11:35] <rvba> strikov: yeah, that's a good workaround.
[11:35] <strikov> rvba: understood, okay, will do
[11:35] <rvba> strikov: ta
[11:35] <jtv> Great, thanks.
[11:35] <jtv> spasiba.
[11:38] <bigjools> rvba: just pushed up to r2140
[11:41] <rvba> Real-world boot images display, for those interested: http://people.canonical.com/~rvb/boot_images_ui.png
[11:41] <jtv> Nice.
[11:41] <rvba> The sorting could be better :)
[11:42] <jtv> I was just writing that I wasn't going to ask about that, because...
[11:42] <jtv> priorities!
[11:42] <jtv> But it's nice to have some visibility at last.
[11:42] <rvba> Yes
[11:48] <jtv> Review needed: https://code.launchpad.net/~jtv/maas/labels-in-new-import-script/+merge/211910
[11:50] <rvba> jtv: I'll take it
[11:50] <jtv> Thanks.
[11:50] <jtv> Disappointingly small, given the trouble it's caused.
[11:50] <jtv> Argh!  I accidentally committed my pserv.yaml changes.
[11:50] <jtv> Let me fix that.
[11:53] <rvba> jtv: why? These changes are welcome.
[11:53] <jtv> Not with the commented-out lines...
[11:54] <jtv> I can add back the actual improvements.
[11:55] <bigjools> who subscribed the team to the int branch?
[11:55] <bigjools> don't do that
[11:56] <jtv> Are we subscribed automatically maybe?
[11:56] <bigjools> dunno
[11:56] <jtv> Because it's owned by the team?
[11:56] <strikov> rvba, jtv, allenap, bigjools, gmb: Workaround to be able to pull Scott's meta (choose armhf in pserv.yaml to avoid pulling tons of images): https://pastebin.canonical.com/106841/
[11:57] <jtv> \o/
[11:57] <rvba> strikov: ta
[11:57] <rvba> This probably deserves a comment in the code though :)
[11:57] <jtv> This isn't for committing — it's for experimenting.
[11:57] <strikov> rvba: that a very dirty workaround :)
[11:57] <rvba> Ah ok.
[11:58] <jtv> Which I think is very empowering.
[11:58] <rvba> I thought we wanted to commit that… until simplestreams gets fixed.
[11:58] <jtv> No, it's just a workaround I requested so we could experiment.
[12:00] <rvba> Reviewer needed (tiny review) https://code.launchpad.net/~rvb/maas/bootimg-ui-sort/+merge/211912
[12:01] <gmb> rvba:
[12:01] <rvba> jtv: you need to manually merge your branch to get it landed on the integration branch.
[12:01] <gmb> rvba: Approved, even :)
[12:01] <rvba> heh, thanks gmb
[12:02]  * gmb -> afk for a few hours; back later.
[12:03] <strikov> jtv, rvba: Scott is fixing this hash issue right now
[12:04] <rvba> strikov: cool
[12:04] <jtv> Even better.
[12:05] <bigjools> rvba: I've been staring at code and re-reading over and over and finally I think I need to accept I am too tired to continue.  I'll catch up tomorrow, don't forget your EOD email :)
[12:06] <bigjools> I'll send one of my own
[12:06] <jtv> nn bigjools
[12:06] <bigjools> nytol
[12:06] <rvba> nn bigjools
[12:26] <jtv> Oh, strikov, one thing I was wondering about: why the strange date format in the name of the snapshot directory?  Why not YYYYMMDD?
[12:26] <jtv> Much better for sorting!
[12:27] <jtv> I noticed because the name looked like a date in 2003.  :)
[12:27] <strikov> jtv: no idea.... that was the last thing i worried about. YYYYMMDD looks really well ;)
[12:27] <jtv> If it doesn't cause any problems, would you mind if I changed it?
[12:28] <strikov> jtv: sure
[12:28] <jtv> Thanks.
[12:31] <strikov> integra
[12:31] <strikov> ouch, sorry
[12:31] <alexpilotti> allenap: ping
[12:32] <jtv> strikov: looks like I got a successful import...  the files are still placeholders.
[12:33] <alexpilotti> hi guys
[12:33] <strikov> jtv: really?
[12:33] <strikov> 1.4G -rw-r--r--  9 root root 1.4G Mar 20 11:49 root-image
[12:33] <alexpilotti> I'm working on the maas latest bbits, and with today's brz pull I have a few issues :-)
[12:36] <strikov-lunch> jtv: I'll return back in 15 mins and do a deeper check
[12:37] <strikov-lunch> jtv: we shouldn't have placeholders there afaik and that looks like a bug
[12:40] <alexpilotti> The first error in the logs is: [Thu Mar 20 14:22:20.513813 2014] [proxy:error] [pid 16550:tid 46944109119232] (111)Connection refused: AH00957: HTTP: attempt to connect to 127.0.0.1:5243 (localhost) failed
[12:40] <alexpilotti> it was working fine with yesterday lates bzr bits, so I wonder if anybody has an idea by any chance :-)
[12:41] <rvba> allenap: can you help alexpilotti please ^
[12:41] <alexpilotti> allenap: tx!
[12:45] <alexpilotti> allenap rvba: a stack trace of a related error: http://paste.openstack.org/show/73903/
[12:49] <alexpilotti> allenap rvba: narrowed down to not being able to start: /home/cloudbase/maas/bin/python /home/cloudbase/maas/bin/maas-region-admin runserver 5243 --settings=maas.demo --noreload
[12:49] <alexpilotti> getting "ImportError: cannot import name close_old_connections"
[13:02] <strikov> rvba, jtv: Scott updated his meta. Now it's signed and contains all the required hashes
[13:02] <strikov> rvba, jtv: checking my script against it right now
[13:32] <allenap> Hi alexpilotti. Looking now.
[13:32] <alexpilotti> allenap: tx
[13:34] <allenap> alexpilotti: The first error implies that the Django app is not up. Try `make status` to check if it’s been started, then look at logs/webapp/current to see what might be the issue.
[13:34] <allenap> The second error the same.
[13:35] <allenap> alexpilotti: Ah, okay, I read a bit more of what you wrote :)
[13:36] <alexpilotti> allenap: webapp is getting continuously respawned
[13:36] <alexpilotti> e.g:
[13:37] <alexpilotti> services/web: up (pid 7565) 125 seconds, normally down
[13:37] <alexpilotti> services/webapp: up (pid 10030) 0 seconds, normally down
[13:37] <alexpilotti> services/web: up (pid 7565) 140 seconds, normally down
[13:37] <alexpilotti> services/webapp: up (pid 10251) 1 seconds, normally down
[13:37] <allenap> alexpilotti: Are you running this on Trusty? Can you check the version of Django you’re using? I wonder if close_old_connections only appears in 1.6.
[13:38] <alexpilotti> yeap trusty, did an apt-get upgrade along with the bzr pull
[13:38] <allenap> alexpilotti: Run: apt-cache policy python-django
[13:38] <alexpilotti> python-django                           1.6.1-2
[13:39] <alexpilotti> python-django:
[13:39] <alexpilotti>   Installed: 1.6.1-2
[13:39] <alexpilotti>   Candidate: 1.6.1-2
[13:39] <alexpilotti>   Version table:
[13:39] <alexpilotti>  *** 1.6.1-2 0
[13:39] <alexpilotti>         500 http://us.archive.ubuntu.com/ubuntu/ trusty/main amd64 Packages
[13:39] <alexpilotti>         100 /var/lib/dpkg/status
[13:39] <allenap> alexpilotti: And: bin/py -c 'import django; print django.__file__'
[13:39] <allenap> In the branch.
[13:39] <alexpilotti> "/usr/local/lib/python2.7/dist-packages/django/__init__.pyc"
[13:40] <allenap> Ah, you’ve got a version of Django in /usr/local that might be stale.
[13:43] <alexpilotti> allenap: checking!
[13:47] <alexpilotti> allenap: good ctach: python -c "import django; print django.get_version()"
[13:47] <alexpilotti> 1.3.1
[13:53] <alexpilotti> removed, everything is good now, tx!
[14:05] <tych0> hi allenap, do you have some time to chat today?
[14:07] <allenap> tych0: Yes! I assume to talk about LXC? Do you want to get this done in time for Trusty?
[14:07] <tych0> allenap: that was the idea
[14:07] <tych0> i might just bag it, though
[14:07] <tych0> you guys don't seem to like the patch
[14:07] <allenap> tych0: Yeah, sorry. I do like the idea though. I guess it’s mainly for testing, right?
[14:07] <tych0> and for use in the cloud installer
[14:08] <tych0> we're already doing this but we have a race condition that this would solve
[14:09] <tych0> anyway, i'm open to suggestions
[14:09] <tych0> but i don't want to waste any more time if it's not going to go in
[14:09] <allenap> tych0: If it can wait a few weeks it would make me happy. I’d like to get it in, but properly. For example, we may want to add a hook to decommission a machine, to avoid that power-off-equals-destroy hack.
[14:10] <tych0> allenap: from my POV we could just power it off and leave it alive, and delete it next time it comes up
[14:10] <tych0> that's basically what maas already does
[14:10] <tych0> i don't mind either way, really
[14:11] <tych0> allenap: waiting a few weeks doesn't really do it for me. if it's not in trusty we can't use it in the cloud installer, so i have much less incentive to fix it.
[14:12] <allenap> tych0: We do have nascent plans to overhaul the transitions a node goes through. Supporting this would be a good use-case. If it can wait, let’s do it properly.
[14:12] <allenap> tych0: Okay, let’s talk at 1430 UTC. Is that okay?
[14:13] <tych0> allenap: that's in 15 minutes or so?
[14:13] <allenap> tych0: Yep.
[14:13] <tych0> allenap: sounds good
[14:14] <allenap> gmb, rvba, jtv: Either of you want to talk about LXC in MAAS too?
[14:14] <allenap> a/Either/Any/
[14:15] <rvba> allenap: I think I'll pass… I'm too deep into the integration work.
[14:32] <allenap> tych0: Problems without hangout?
[14:32] <allenap> Or with even.
[14:35] <allenap> tych0: Try https://talky.io/maas
[14:36] <tych0> allenap: joining now