| wallyworld | thumper: you around? | 00:03 |
|---|---|---|
| bradm | anyone know much about neutron? having issues with a juju deployed openstack | 00:07 |
| wallyworld | bradm, mgz knows most about neutron but he's probably asleep | 00:14 |
| bradm | wallyworld: I seem to be on the wrong side of the world to be dealing with neutron, everyone who seems to know it is asleep | 00:15 |
| wallyworld | yeah :-( | 00:16 |
| thumper | wallyworld: hi | 00:25 |
| wallyworld | gday | 00:25 |
| wallyworld | got time for a chat? | 00:25 |
| thumper | yup | 00:26 |
| thumper | ish | 00:26 |
| wallyworld | it will be quick, just a heads up from the standup last night | 00:27 |
| wallyworld | https://plus.google.com/hangouts/_/calendar/bWFyay5yYW1tLWNocmlzdGVuc2VuQGNhbm9uaWNhbC5jb20.sbtpoheo4q7i7atbvk9gtnb3cc?authuser=1 | 00:28 |
| davecheney | 2014-02-26 01:11:42 ERROR juju runner.go:220 worker: exited "environ-provisioner": failed to process updated machines: error executing "lxc-ls": wait: no child processes | 01:12 |
| davecheney | 2014-02-26 01:11:43 ERROR juju.container.lxc lxc.go:166 failed getting all instances: error executing "lxc-ls": wait: no child processes | 01:12 |
| davecheney | ^ local provider peeps | 01:12 |
| davecheney | any ideas ? | 01:12 |
| axw | can't say I've seen that. busted packages again? :( | 01:14 |
| davecheney | is lxc-ls bailing before the wait occurs ? | 01:17 |
| axw | ah, I read it as an error coming back from lxc-ls | 01:19 |
| davecheney | ubuntu@winton-02:~$ lxc-ls | 01:19 |
| davecheney | ubuntu@winton-02:~$ echo $? | 01:19 |
| davecheney | 0 | 01:19 |
| axw | not sure. | 01:19 |
| davecheney | weird | 01:19 |
| axw | try lxc-ls -1 | 01:19 |
| axw | (one, not ell) | 01:19 |
| axw | and again as root | 01:20 |
| davecheney | ubuntu@winton-02:~$ sudo lxc-ls | 01:20 |
| davecheney | dave foo ken ryu | 01:20 |
| davecheney | interesting | 01:20 |
| davecheney | ubuntu@winton-02:~$ ps auxww | grep jujud | 01:20 |
| davecheney | root 24736 3.0 7.5 8262756 621576 ? Ssl Feb22 174:19 /home/ubuntu/.juju/local/tools/machine-0/jujud machine --data-dir /home/ubuntu/.juju/local --machine-id 0 --debug | 01:20 |
| davecheney | but the machine agent is running as root | 01:20 |
| axw | does "sudo lxc-ls -1" do anything different? | 01:21 |
| axw | doubtful.. | 01:21 |
| axw | davecheney: the golxc code is just doing exec.Command("lxc-ls", "-1").CombinedOutput() | 01:23 |
| axw | so... *shrug* | 01:23 |
| davecheney | ubuntu@winton-02:~$ sudo lxc-ls -1 | 01:23 |
| davecheney | dave | 01:23 |
| davecheney | foo | 01:23 |
| davecheney | ken | 01:23 |
| davecheney | ryu | 01:23 |
| davecheney | same, but different | 01:23 |
| davecheney | screw it | 01:23 |
| axw | yeah -1 just means one per line, didn't think it'd matter | 01:24 |
| davecheney | gonna destroy and start again | 01:24 |
| thumper | waigani: ping - standup | 01:31 |
| davecheney | + /home/ubuntu/bin/juju destroy-environment -y local | 01:50 |
| davecheney | + /home/ubuntu/bin/juju bootstrap | 01:50 |
| davecheney | ERROR cannot use 37017 as state port, already in use | 01:50 |
| davecheney | sigh | 01:50 |
| davecheney | bloody hell | 01:51 |
| davecheney | what ? | 01:51 |
| davecheney | there is nothing listening on that port .... | 01:51 |
| davecheney | ERROR cannot use 37017 as state port, already in use | 01:52 |
| davecheney | ubuntu@winton-02:~$ ss | grep 37017 | 01:52 |
| davecheney | ubuntu@winton-02:~$ | 01:52 |
| mwhudson | davecheney: i don't think that's a good way to check | 01:54 |
| mwhudson | you want ss -nl i think? | 01:54 |
| davecheney | oh | 01:59 |
| davecheney | serves me right for using hipster tools | 01:59 |
| davecheney | i'll go back netstat -anp next time | 01:59 |
| davecheney | ubuntu@winton-02:~$ sudo !! | 01:59 |
| davecheney | sudo netstat -anp | grep 37 | 01:59 |
| davecheney | tcp 0 0 0.0.0.0:37017 0.0.0.0:* LISTEN 24711/mongod | 01:59 |
| davecheney | intersting | 02:00 |
| davecheney | i wonder why it didn't die | 02:00 |
| === edu-afk is now known as edamato | ||
| === mwhudson is now known as zz_mwhudson | ||
| === zz_mwhudson is now known as mwhudson | ||
| thumper | axw, wallyworld, waigani: we'll just use this one... https://plus.google.com/hangouts/_/calendar/dGltLnBlbmhleUBjYW5vbmljYWwuY29t.kieq2g96khnegtjsretb0l4clo | 02:29 |
| wallyworld | thumper: says not allowed to join | 02:31 |
| thumper | pfft | 02:31 |
| axw | same | 02:31 |
| * thumper sighs | 02:31 | |
| thumper | this was the one from the package review | 02:31 |
| thumper | you are now invited | 02:32 |
| === edamato is now known as edu-afk | ||
| jam | thumper: poke, are you still online? | 03:03 |
| thumper | jam: yes, am with the team talking timings :) | 03:03 |
| jam | thumper: I'm available for video chat for a bit if you want | 03:06 |
| thumper | jam: available now | 03:20 |
| jam | thumper: starting up G+ hangout now | 03:31 |
| thumper | kk | 03:31 |
| jam | thumper: https://plus.google.com/hangouts/_/7ecpikfgr7oi6hsg74ds6jdcl0 | 03:31 |
| === mwhudson is now known as zz_mwhudson | ||
| bigjools | thumper: seriously lol-ing at your twitter conversation this morning | 04:45 |
| axw | oh wtf | 06:23 |
| * axw stops wtf'ing | 06:25 | |
| axw | the bot marked the wrong MP as merged | 06:25 |
| jam | axw: where? | 07:01 |
| jam | I can't say that I've seen that before | 07:02 |
| jam | the bot doesn't actually do it | 07:02 |
| jam | Launchpad detects when the tip revision has been merged into trunk | 07:02 |
| jam | so it is generally accurate | 07:02 |
| axw | jam: https://code.launchpad.net/~axwalk/juju-core/lp1281071-rsyslog-tls/+merge/207889 | 07:02 |
| axw | what I did was I created another branch, merged that one in but removed all the stuff I didn't care about | 07:02 |
| axw | so LP saw the merged revisions | 07:03 |
| jam | axw: the above branch is "andrew.wilkins@canonical.com-20140225140210-4206fc8pbnfkpb46" which is rev 2349.2.5 in trunk | 07:03 |
| jam | so that MP was *merged*, just from something else | 07:03 |
| axw | yeah, makes sense - I just haven't done that before and was a bit surprised | 07:05 |
| jam | be careful when clicking near tabs :) | 07:05 |
| axw | :) | 07:06 |
| axw | <axw> yeah, makes sense - I just haven't done that before and was a bit surprised | 07:06 |
| axw | jam: yes, the rev is there, I just took out most of what's in that MP | 07:06 |
| dimitern | fwereade, wrtp, i'd appreciate a review and some direction on https://codereview.appspot.com/66540044 | 09:25 |
| wrtp | dimitern: looking | 09:29 |
| wrtp | dimitern: reviewed | 09:51 |
| dimitern | wrtp, ta! | 09:52 |
| mattyw | dimitern, your review here: https://codereview.appspot.com/67870043/ do you mean I need to implement the new status code in FullStatus only - and not in Status? | 09:56 |
| dimitern | mattyw, FullStatus only - Status uses FS internally, when needed (1.16 compat mode) | 09:58 |
| dimitern | fwereade, wrtp, so, it seems we'll need to define the next agent conf version - should it be 1.17 or 1.18? I'm thinking of adding Version, DataDir, LogDir, possibly Jobs as fields | 09:59 |
| === wrtp is now known as rogpeppe | ||
| rogpeppe | dimitern: i don't think it makes sense to define formats for dev versions | 10:00 |
| mattyw | dimitern, ok thanks | 10:00 |
| rogpeppe | dimitern: so i'd suggest 1.18 | 10:00 |
| dimitern | rogpeppe, ok, and in that version we'll use a single file + version, no format separately | 10:02 |
| rogpeppe | dimitern: yeah | 10:02 |
| dimitern | rogpeppe, can you think of any other fields we might need to define except the ones above? | 10:02 |
| rogpeppe | dimitern: what is "Version" there? the agent configuration version? | 10:03 |
| dimitern | rogpeppe, ok Format then | 10:05 |
| rogpeppe | dimitern: what does it hold? | 10:05 |
| dimitern | rogpeppe, the agent conf format version | 10:07 |
| rogpeppe | dimitern: i'm not sure we want to expose that outside of the agent package | 10:10 |
| rogpeppe | dimitern: and i don't think it should be a regular field | 10:10 |
| rogpeppe | dimitern: i suggest that we store the version on a line of its own before any of the other conf data | 10:10 |
| dimitern | rogpeppe, expand? | 10:10 |
| rogpeppe | dimitern: then we can read the version and use it to decide how to parse the rest of the config data | 10:11 |
| rogpeppe | dimitern: which gives us freedom to change file formats in the future if we want (for example to move away from YAML) | 10:11 |
| * rogpeppe has just got his laptop back through the post | 10:12 | |
| rogpeppe | and they've cleaned it all up - it doesn't look like my scruffy laptop any more :-) | 10:12 |
| rogpeppe | let's see if it works... | 10:13 |
| rogpeppe | ha, they replaced the keyboard! | 10:13 |
| jam | rogpeppe: I'm glad you got it back in time, congrats | 10:17 |
| dimitern | rogpeppe, \o/ | 10:18 |
| dimitern | rogpeppe, ok, that sgtm | 10:18 |
| rogpeppe1 | jam: yay! back on the thinkpad! | 10:20 |
| === rogpeppe1 is now known as rogpeppe | ||
| === rogpeppe1 is now known as rogpeppe | ||
| jam | rogpeppe: fwereade, dimitern: standup? | 10:46 |
| jam | fwereade: poke ? | 10:50 |
| fwereade | jam, sorry, couple of mins | 10:51 |
| jam | mgz: poke ? | 10:58 |
| mgz | jam: ta | 10:59 |
| jam | fwereade: one more quick standup ping? | 11:08 |
| rogpeppe | fwereade: https://codereview.appspot.com/68650044 | 11:48 |
| jam | rogpeppe: https://bugs.launchpad.net/juju-core/+bug/1254577 is still marked Critical and not Fix Committed. I thought you landed code related to that? | 11:51 |
| rogpeppe | jam: marked as "fix committed" | 11:52 |
| jam | rogpeppe: do you know if that got released in 1.17.3 | 11:52 |
| jam | ? | 11:52 |
| rogpeppe | jam: i certainly hope so. let me check. | 11:53 |
| rogpeppe | jam: well, the fix was merged 3 weeks ago, so i should certainly hope so | 11:55 |
| jam | does anyone have the link to the current CI runs? I seem to have an old IP address that is no longer valid | 11:57 |
| jam | sinzui: ^^ | 11:57 |
| natefinch | jam: FYI I have an old IP too, not sure what happened. Sinzui is likely still asleep, but when he responds, I'll email out the new IP. | 12:13 |
| fwereade | frankban, dimitern: does charm get handle symlinks at all? | 12:21 |
| dimitern | fwereade, because it uses a zip reader internally, i guess it should | 12:22 |
| fwereade | frankban, dimitern: it looks like it's just sending the target path and I'm not convinced that's useful | 12:22 |
| frankban | fwereade: I guess it depends on how they are handled by zip.ReadCloser | 12:23 |
| dimitern | frankban, fwereade, but a test with a zip + symlinks is definitely useful | 12:24 |
| fwereade | frankban, dimitern: ok, what do you think we should be doing? sending the actual content of the target? | 12:24 |
| frankban | fwereade: makes sense | 12:25 |
| dimitern | fwereade, frankban, that sounds reasonable | 12:25 |
| frankban | fwereade, dimitern: trying with a symlink | 12:25 |
| fwereade | dimitern, frankban: ok, cheers, I'll see what I can do | 12:25 |
| fwereade | dimitern, frankban: also, would it be *really* annoying if I changed the filelist stuff to return directories as well? I freely admit this is more rooted in convenience than any solid design principles | 12:28 |
| fwereade | dimitern, frankban: hmm, I think we may need special handling for the revision file too | 12:29 |
| frankban | fwereade: not annoying for me | 12:29 |
| fwereade | frankban, cool, basically I'd like to just return the result of the imminently-available Bundle.Manifest method | 12:30 |
| frankban | fwereade: re: revision file, is it special? | 12:30 |
| fwereade | frankban, its presence is not guaranteed in the zip file, but it is always returned by Manifest because we do always write one out | 12:31 |
| frankban | fwereade: oh, so you'll instantiate a bundle and it will have a Manifest returning the file list? | 12:31 |
| fwereade | frankban, it'll *almost* certainly always be there, but it's that "almost" that's the issue ;) | 12:31 |
| fwereade | frankban, yeah, I think so -- is a ReadBundle unreasonable at that point? | 12:31 |
| frankban | fwereade: OIC, if it's not there there is no problem from our perspective, AFAICT we are not interested in the revision file for that API. but I agree, if it's in the list, we must also return its contents | 12:32 |
| frankban | fwereade: in my previous impl (before requirements changes) I used ReadBundle to and ExpandTo to extract the files, so, it makes sense. I wonder if at that point a bundle.GetFile would be reasonable too. | 12:34 |
| * fwereade starts wondering about a Content(path) method | 12:36 | |
| fwereade | frankban, yeah, I'm just having trouble integrating the symlink-resolution stuff in my current model | 12:37 |
| fwereade | frankban, how do we currently behave if you ask for a path to a dir? | 12:41 |
| fwereade | frankban, ah 403 cool | 12:41 |
| frankban | fwereade: forbidden | 12:41 |
| frankban | yeah | 12:41 |
| fwereade | frankban, ok, I think that works for me | 12:41 |
| fwereade | frankban, dimitern: can we do a really quick hangout? | 12:42 |
| frankban | fwereade: sure | 12:43 |
| dimitern | fwereade, frankban, sure | 12:43 |
| fwereade | dimitern, frankban: https://plus.google.com/hangouts/_/7ecpjumv8629dqrig0jdegier8?hl=en | 12:43 |
| rogpeppe | fwereade: you've got a review: https://codereview.appspot.com/69110043/ | 13:19 |
| sinzui | jam, natefinch http://162.213.35.54:8080/ and it appears ill (and we stood it up 2 days ago too) | 13:21 |
| sinzui | oh, no route to host | 13:21 |
| jam | sinzui: k, so that I can't get to it is probably because it isn't there :) | 13:21 |
| natefinch | heh | 13:22 |
| === psivaa is now known as psivaa-lunch | ||
| rogpeppe | dimitern: i've responded to your review | 14:12 |
| rogpeppe | dimitern: i'd appreciate the rest, if you have some time | 14:13 |
| dimitern | rogpeppe, thanks, i'll look into it a bit later | 14:14 |
| rogpeppe | dimitern: ta | 14:14 |
| rogpeppe | natefinch, fwereade: your reviews appreciated too | 14:14 |
| === psivaa-lunch is now known as psivaa | ||
| niemeyer | rogpeppe: Hey, just a ping to let you know your branch got merged, and the tests that were failing due to the panic formatting changes in recent Go versions have also been fixed | 14:51 |
| rogpeppe | niemeyer: great, thanks. i was planning to get around to merging it, but my laptop died which was a distraction | 14:52 |
| rogpeppe | niemeyer: i'll update the juju-core deps file | 14:52 |
| niemeyer | rogpeppe: Thanks | 14:52 |
| dimitern | rogpeppe, what are these cryptic 0v 1p in member definitions | 14:59 |
| rogpeppe | dimitern: see the definition of mkMembers and mkStatuses | 14:59 |
| dimitern | rogpeppe, right, thanks | 15:00 |
| niemeyer | mramm, fwereade: Are we having that call today? | 15:00 |
| mramm | I'm in the hangout | 15:01 |
| mramm | if you want to have it | 15:01 |
| mramm | but I don't have any items for today's agenda | 15:01 |
| mramm | so I am also fine with skipping today | 15:02 |
| niemeyer | mramm: I'm joining | 15:03 |
| natefinch | rogpeppe: resuming the review now | 15:21 |
| rogpeppe | natefinch: thanks! | 15:21 |
| rogpeppe | lunch | 15:33 |
| sinzui | natefinch, rogpeppe , is someone adding support to install juju-mongodb when it is available? | 15:46 |
| rogpeppe | sinzui: i think that's the plan, but i don't know its current status | 15:46 |
| sinzui | rogpeppe, thank you | 15:47 |
| dimitern | rogpeppe, reviewed | 15:50 |
| rogpeppe | dimitern: ta! | 15:50 |
| natefinch | sinzui: it is planned but I don't think anyone is assigned. I did the other part, but someone else should probably do this part, since I am way behind on HA stuff. | 15:50 |
| sinzui | natefinch, fab. I think I will work to release juju 1.17.4 as it is for people working on trusty and other archs | 15:51 |
| dimitern | rogpeppe, fwereade, so with the introduction of agent conf format 1.18 we should drop 1.12 format support, right? | 16:03 |
| fwereade | rogpeppe, would you explain the windows 022 thing a bit more? | 16:03 |
| rogpeppe | fwereade: i think so | 16:03 |
| rogpeppe | dimitern: i think so | 16:03 |
| dimitern | ok :) | 16:03 |
| fwereade | dimitern, hmm, I have a nasty feeling that 1.12 might try to upgrade arbitrarily far | 16:03 |
| rogpeppe | fwereade: so, i'm not convinced that we'll get sane-looking perms from windows-generated zip files | 16:04 |
| dimitern | fwereade, so? | 16:04 |
| dimitern | fwereade, it will fail | 16:04 |
| rogpeppe | fwereade: from previous experience with unix-like perms on windows | 16:04 |
| rogpeppe | fwereade: natefinch may know more of the expected behaviour | 16:04 |
| fwereade | dimitern, so I'm not sure that we can guarantee that 1.18 won't start up with a 1.12 conf | 16:04 |
| fwereade | dimitern, it may just be a matter of shouting about it in the release notes though | 16:05 |
| fwereade | rogpeppe, natefinch: in particular I don't quite get what `&=022` would do, maybe because I'm not sure where you mean me to apply it | 16:06 |
| dimitern | fwereade, well, we can't guarantee compatibility 16 releases back anyway, so it should just fail in some sensible way | 16:06 |
| rogpeppe | fwereade: it would clear group and other write permissions from all files and directories | 16:07 |
| rogpeppe | fwereade: sorry, it should have been &^022 | 16:08 |
| rogpeppe | fwereade: another possibility is to use ((p&7) | (p&7)<<3 | (p&7)<<6)) & ^022 | 16:09 |
| rogpeppe | fwereade: i.e. ignore group and other modes, don't allow any group/user write perms, and use r and x bits from user perms for group and other | 16:10 |
| rogpeppe | fwereade: but that means it's not possible to make a non-group or other -readable file | 16:10 |
| natefinch | fwereade, rogpeppe: catching up.... perms on windows are way different. Not 100% sure on how Go translates permission bits to windows | 16:26 |
| rogpeppe | natefinch: yeah. | 16:26 |
| rogpeppe | natefinch: what do you think windows-created zip files will report for permissions? | 16:26 |
| rogpeppe | natefinch: perhaps you could create a zip file under windows, with a variety of permissions, and we could find out | 16:27 |
| natefinch | rogpeppe: yeah, that';s what I was thinking. I don't know what linux defaults to when it sees a windows file that doesn't have permissions like that set. I'll try it out, easy enough | 16:28 |
| rogpeppe | fwereade: one thought is that if we're unpacking the charm under unix, it's pretty unlikely that the charm was zipped under windows | 16:28 |
| rogpeppe | natefinch: i did some of the go zip perms code, ages ago, but i can't remember much about it and it's probably changed since i did it | 16:28 |
| rogpeppe | natefinch: i think the trauma of reading the zip spec must've blanked my memory :-] | 16:28 |
| natefinch | rogpeppe: hahah | 16:29 |
| fwereade | natefinch, rogpeppe: sorry, multitasking fail | 16:34 |
| natefinch | rogpeppe: by default it's 0664 | 16:36 |
| rogpeppe | natefinch: which default? | 16:36 |
| natefinch | -rw-rw-r-- for those of us less binarily inclined | 16:36 |
| natefinch | rogpeppe: if I just create a zip without specifying permissions | 16:36 |
| rogpeppe | natefinch: is that Go's behaviour or windows' ? | 16:36 |
| natefinch | rogpeppe: windows, sorry | 16:37 |
| rogpeppe | natefinch: have you tried unpacking that using Go, to make sure | 16:37 |
| natefinch | rogpeppe: not yet, I'll do that | 16:37 |
| natefinch | rogpeppe: unpacking on linux, I presume? | 16:38 |
| rogpeppe | natefinch: it shouldn't make any difference | 16:39 |
| rogpeppe | natefinch: you don't actually need to create the file | 16:39 |
| fwereade | rogpeppe, natefinch: fwiw we *do* make sure that hooks are executable when we unpack *anyway* and I suspect this hits most of the cases -- 644/744 should be enough for anyone, right? | 16:39 |
| rogpeppe | natefinch: just see what perms are reported | 16:39 |
| fwereade | rogpeppe, natefinch: indeed assuming that *is* what's reported | 16:39 |
| rogpeppe | fwereade: i'd go for 755 rather than 744 | 16:40 |
| rogpeppe | fwereade: but yeah, i don't think there's a huge issue here | 16:40 |
| rogpeppe | fwereade: it's worth going into it with our eyes open though | 16:40 |
| rogpeppe | fwereade: (i have definitely hit problems in this area in a previous lifetime) | 16:40 |
| fwereade | rogpeppe, natefinch: I'm not *really* expecting we'll get a lot of people writing charms on windows anyway tbh | 16:41 |
| fwereade | rogpeppe, natefinch: drag a charm archive into a browser on windows? plausible | 16:42 |
| fwereade | rogpeppe, natefinch: zip up a subtly wrong dir and drag it into a browser on linux? also plausible | 16:42 |
| fwereade | rogpeppe, natefinch: "develop" a charm sight unseen on windows, and deploy it to a real ubuntu cloud straight off? seems less likely | 16:43 |
| rogpeppe | fwereade: it's possible but i think that the resulting issues probably won't be *too* bad | 16:43 |
| rogpeppe | fwereade: exactly | 16:43 |
| rogpeppe | fwereade: one other thing occurs to me: the perms you'll get when creating a file with os.OpenFile may well be different from the perms you get when doing a chmod | 16:45 |
| rogpeppe | fwereade: because of umask | 16:45 |
| rogpeppe | fwereade: to be consistent, perhaps we should AND all the perms with the current umask before doing anything | 16:46 |
| fwereade | &^umask, right? | 16:48 |
| natefinch | rogpeppe, fwereade: btw the unzipped files w/ go are created with 0664, too. | 16:50 |
| rogpeppe | fwereade: yeah | 16:51 |
| rogpeppe | natefinch: that's not a great default, but umask should sort it out | 16:51 |
| fwereade | natefinch, ok, that seems probably good enough to me? | 16:54 |
| fwereade | rogpeppe, added v brief notes on https://codereview.appspot.com/68650044/ -- I'll be looking at it some more | 16:54 |
| fwereade | rogpeppe, but it is explicitly *not* nlgtmed, so please do not take my need to convince myself further as a roadblock | 16:55 |
| rogpeppe | fwereade: i'd be interested to know what kind of thing you're thinking about when you suggest a "simpler and dumber model" | 16:59 |
| rogpeppe | fwereade: i agree that the current mocking setup is somewhat experimental, but i *think* it beats trying to get the real State to jump through all those hoops | 17:02 |
| fwereade | rogpeppe, yeah, definitely agreed | 17:03 |
| rogpeppe | fwereade: the IsNotFound cases are the only place in the code we don't have 100% coverage. that's mainly because i haven't yet thought of a decent way to "wait for nothing to happen" | 17:06 |
| rogpeppe | fwereade: i'd also be interested to know where you found it difficult to follow the code, so i can add an appropriate comment or two | 17:07 |
| fwereade | rogpeppe, I'll hopefully have some better comments later today | 17:23 |
| rogpeppe | fwereade: i've just approved the MP BTW, but i'll take on board any further comments and integrate them into the next branch | 17:24 |
| fwereade | rogpeppe, I do find myself wondering whether we actually kinda want to unpack charms with the permissions in the zip file, umask be damned | 17:24 |
| fwereade | rogpeppe, thanks | 17:24 |
| rogpeppe | fwereade: i'm not sure | 17:24 |
| rogpeppe | fwereade: an easy way to do that is to set umask to 0 | 17:24 |
| rogpeppe | fwereade: umask is kinda dumb anyway | 17:25 |
| rogpeppe | fwereade: i preferred the plan 9 way of doing things, where the default perms were taken from the parent dir | 17:25 |
| fwereade | rogpeppe, heh, but then we do want to keep using the default umask for hook execution I think | 17:25 |
| * fwereade rather agrees tere | 17:25 | |
| rogpeppe | fwereade: i don't know. what do you think the expectation of umask is for charm authors? | 17:26 |
| fwereade | rogpeppe, there were bugs about it in the past | 17:27 |
| fwereade | rogpeppe, iirc the expectation is "execute hooks with the system umask" | 17:27 |
| rogpeppe | fwereade: the other thing that i didn't remark on in the review, is what happens if you have a directory with non-writable permissions, but it contains some files? | 17:28 |
| rogpeppe | fwereade: i *think* that in general it's a moot point because zip files don't usually mention directories explicitly | 17:28 |
| rogpeppe | fwereade: but it can be an issue | 17:28 |
| fwereade | rogpeppe, yeah, my thinking was mainly that unwritable things are unlikely to come up in practice, and when they do we have ErrConflict | 17:32 |
| fwereade | rogpeppe, because I certainly don't know what the point of having such things would be | 17:32 |
| rogpeppe | fwereade: alternatively, make sure that all directories are at least 700 | 17:32 |
| rogpeppe | fwereade: me neither, but it's difficult to pre-guess | 17:33 |
| fwereade | rogpeppe, and thus am unqualified to impose expectations | 17:33 |
| ev | hiya. Is ssh forwarding broken? http://paste.ubuntu.com/7000920/ | 17:35 |
| ev | that's on 1.17.3-saucy-amd64 | 17:35 |
| rogpeppe | ev: hmm, it would be interesting to find out what actual command line is being executed there | 17:37 |
| ev | rogpeppe: hm? Isn't that in the pastebin? | 17:37 |
| mgz | is that just bug 1281577 | 17:37 |
| _mup_ | Bug #1281577: juju 1.17.2 doesn't pass command line options to ssh <docs> <landscape> <regression> <ui> <juju-core:Fix Committed by dimitern> <https://launchpad.net/bugs/1281577> | 17:37 |
| ev | ah, that's the badger | 17:37 |
| ev | thanks guys | 17:37 |
| rogpeppe | ev: no, sorry, i mean the ssh comment | 17:37 |
| rogpeppe | mgz: thanks | 17:38 |
| mgz | ev: sinzui is doing 11.17.4 today which has the fix | 17:38 |
| mgz | -1 | 17:38 |
| ev | whoop! | 17:41 |
| natefinch | rogpeppe, fwereade: btw, it looks like zip files don't actually include the directory information, just implicitly through paths on files (at least the ones I created on windows 7). | 17:42 |
| fwereade | natefinch, would you mail me one please? | 17:43 |
| rogpeppe | natefinch: that's what i meant by "zip files don't usually mention directories explicitly" | 17:43 |
| rogpeppe | natefinch: i think it's possible for them to do so though | 17:43 |
| rogpeppe | natefinch: i *think* it's possible to have empty dirs in a zip file | 17:44 |
| natefinch | rogpeppe: windows won't let you do it : [Window Title] | 17:45 |
| natefinch | Compressed (zipped) Folders | 17:45 |
| natefinch | [Content] | 17:45 |
| natefinch | Windows was unable to add one or more empty directories to the Compressed (zipped) Folder. | 17:45 |
| natefinch | [OK] | 17:45 |
| rogpeppe | natefinch: ha ha | 17:45 |
| rogpeppe | natefinch: no one likes those poor empty directories | 17:45 |
| natefinch | rogpeppe: I'd just been testing that | 17:45 |
| rogpeppe | natefinch: apart from bzr, which is being phased out | 17:45 |
| natefinch | rogpeppe: yeah, it's kinda sad, because they can have value. | 17:45 |
| rogpeppe | natefinch: they're really important sometimes | 17:46 |
| natefinch | rogpeppe: if for nothing else than "here's where you should put X stuff" | 17:46 |
| mgz | natefinch: certainly tools *exist* on windows that let you add empty dirs to zips... | 17:47 |
| mgz | whether they're what people use is another matter | 17:47 |
| natefinch | mgz: yeah, probably 7zip etc would allow it. Certainly we should support it if it's possible. | 17:47 |
| mgz | 7zip does, just tested | 17:47 |
| natefinch | mgz: good to know. | 17:50 |
| natefinch | rogpeppe: we should probably talk HA next steps before your EOD, or jam will have my head for not putting up some tasks on kanban | 17:54 |
| rogpeppe | natefinch: i'm just working on that, but let's have a chat about it | 17:55 |
| rogpeppe | natefinch: in fact, give me 10 mins to finish my thoughts here | 17:55 |
| natefinch | rogpeppe: sure thing' | 17:55 |
| sinzui | mgz, ev : I hope to do a 1.17.4 release tomorrow. nova/canonistack lost the ci machine abotu 12 hours. We had to stand up another one as our first priority | 18:06 |
| mgz | sinzui: eck... we need a less overcommitted region probably :0 | 18:07 |
| sinzui | mgz, It happened last week too. We will move to another region or cloud | 18:08 |
| ev | sinzui: *nods* I've just sent instructions to my team on using juju from trunk for this | 18:12 |
| rogpeppe | natefinch: https://plus.google.com/hangouts/_/calendar/am9obi5tZWluZWxAY2Fub25pY2FsLmNvbQ.mf0d8r5pfb44m16v9b2n5i29ig?authuser=1 | 18:12 |
| ev | HP has hilariously set really low limits on not only security groups, but security group rules | 18:12 |
| ev | so wooo tunnels | 18:12 |
| ev | I don't suppose I could convince someone to update the simplestreams stuff to use the new US/West HP Cloud region? | 18:13 |
| ev | we've uploaded the right thing to a public bucket on one of our HPCloud accounts for now | 18:13 |
| sinzui | ev, Juju-CI just built UNRELASED debs for trusty, saucy, and precise. This is the release candidate we are testing: http://162.213.35.54:8080/job/publish-revision/ | 18:14 |
| sinzui | if ev, this is the actual release tarball we will use on success: http://162.213.35.54:8080/job/build-revision/ | 18:14 |
| ev | ( https://region-a.geo-1.objects.hpcloudsvc.com/v1/11289530460295/images and https://region-a.geo-1.objects.hpcloudsvc.com/v1/11289530460295/tools ) | 18:15 |
| ev | sinzui: *nods* thanks! | 18:15 |
| === psivaa is now known as psivaa-afk | ||
| * rogpeppe is done for the day | 19:20 | |
| rogpeppe | g'night all | 19:20 |
| rogpeppe | fwereade: there are now some kanban cards for HA, BTW | 19:21 |
| thumper | o/ | 19:51 |
| === zz_mwhudson is now known as mwhudson | ||
| === smoser` is now known as smoser | ||
| === dstroppa_ is now known as dstroppa | ||
| === hatch_ is now known as hatch | ||
| thumper | waigani: https://bugs.launchpad.net/juju-core/+bug/1281394 | 21:59 |
| _mup_ | Bug #1281394: uniter failed to run non-existant config-changed hook <regression> <juju-core:Triaged by thumper> <https://launchpad.net/bugs/1281394> | 21:59 |
| thumper | wallyworld: just for you: http://pastebin.ubuntu.com/7002116/ | 22:10 |
| wallyworld | looking | 22:10 |
| wallyworld | is this trunk? | 22:11 |
| thumper | aye | 22:11 |
| thumper | pulled trunk, started a local provider | 22:11 |
| wallyworld | hmmm. worked for me before :-( | 22:11 |
| wallyworld | and CI | 22:11 |
| wallyworld | ok, i'll have to look | 22:12 |
| thumper | and also deployed ubuntu charm | 22:12 |
| thumper | that is all | 22:12 |
| thumper | well, it is up and running | 22:12 |
| thumper | but the worker is bouncing | 22:12 |
| wallyworld | sigh, yeah will look | 22:12 |
| wallyworld | but i'm in a good mood - got support for IDE based debugging :-) | 22:12 |
| wallyworld | niiiiiice | 22:13 |
| wallyworld | no more fmt.Printf() to see wtf is going on | 22:13 |
| thumper | haha | 22:21 |
| thumper | nice | 22:21 |
| waigani | Tim just shot me | 22:21 |
| thumper | nerf | 22:22 |
| wallyworld | thumper: what juju version were you upgrading local provider env from? | 22:34 |
| thumper | I wasn't upgrading | 22:34 |
| thumper | I was just starting it | 22:34 |
| wallyworld | hmmm. ok | 22:35 |
| wallyworld | thumper: since you're ocr :-) could you +1 this one roger looked at - i've responded to changes but he didn't get to +1 yesterday. there's also https://codereview.appspot.com/68990043/ :-D | 22:38 |
| thumper | i'm ocr? | 22:38 |
| wallyworld | it says so on the calendar :-) | 22:38 |
| * thumper sighs | 22:38 | |
| thumper | suppose I am then... | 22:38 |
| thumper | yes, I'll look soonish | 22:38 |
| wallyworld | yep :-) | 22:39 |
| wallyworld | no hurry | 22:39 |
| wallyworld | i have 4 branches stacked up i need to land | 22:39 |
| wallyworld | so just keen to clear the queue sometime today or whenever | 22:39 |
| thumper | yeah yeah | 22:40 |
| * thumper puts wallyworld on the TODO soonish stack | 22:41 | |
| wallyworld | no hurry | 22:45 |
| === thumper is now known as thumper-gym | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!