=== rangerpb is now known as rangerpbzzzz | ||
=== sambetts|afk is now known as sambetts | ||
=== rangerpbzzzz is now known as rangerpb | ||
=== natorious_ is now known as natorious | ||
blackboxsw | hey smoser, so where do things stand with the current SRU process? I'm trying to find an sru bug related to the work. | 16:28 |
---|---|---|
smoser | blackboxsw, ok.. collecting some info | 16:30 |
blackboxsw | good deal, thanks. | 16:30 |
smoser | typing https://public.etherpad-mozilla.org/p/cloud-init-sru-info | 16:31 |
smoser | blackboxsw, so i guess the easiest thing to do is just start doing it... we can walk down the list and pick one by one. if you dont understand the test case / verification that i put in for anything, then you can either | 16:35 |
smoser | a.) ask questions | 16:35 |
smoser | b.) go to the next one | 16:35 |
smoser | (until you run out and are forced to 'a') | 16:36 |
blackboxsw | yeah that sounds like a good loop. expect the a's | 16:37 |
* blackboxsw starts at the top | 16:39 | |
blackboxsw | https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/1676460 xenial validation | 16:39 |
smoser | blackboxsw, many of the bugs reference 'lxc-proposed-snapshot' | 16:40 |
smoser | that one did not. it would be good to make yourself a lxc with proposed cloud-init in it. | 16:40 |
smoser | see http://pad.lv/1674766 for an example. | 16:41 |
blackboxsw | cool will do | 16:42 |
smoser | that requires 'mount-image-callback' (cloud-utils) .. and it requires lxc1 for 'lxcnsuserexec' | 16:44 |
smoser | err... lxc-nsuserexec | 16:44 |
smoser | so i just ran | 16:47 |
smoser | ./bin/lxc-proposed-snapshot -vvv xenial --proposed xenial-proposed --publish | 16:47 |
smoser | which means i can now: | 16:48 |
smoser | lxc launch xenial-proposed xp1 | 16:48 |
smoser | and it starts with cloud-init inside from -proposed | 16:48 |
blackboxsw | hrm /home/csmith/src/server/lxd:xenial-proposed-11791681: not a file re-reading the script intent | 16:52 |
blackboxsw | and I have lxc1 & cloud-utils | 16:53 |
blackboxsw | per http://pastebin.ubuntu.com/24375059/ | 16:53 |
smoser | newer cloud-utils needed. :-(. that support i think went into zesty. | 16:54 |
blackboxsw | updating... :/ | 16:55 |
smoser | if you want, you can probably just grab it | 16:55 |
smoser | its a simplish script | 16:55 |
smoser | and put it in $PATH | 16:55 |
blackboxsw | yeah might just do that while waiting on my desktop upgrade. | 16:56 |
smoser | yikes. you are brave. | 17:01 |
smoser | (i run the development release but most do not :) | 17:01 |
smoser | the other option is doing this on an instance somewhere. that is why lxd is nice. | 17:01 |
smoser | blackboxsw, http://bazaar.launchpad.net/~cloud-utils-dev/cloud-utils/trunk/view/head:/bin/mount-image-callback | 17:02 |
blackboxsw | ok, looks better thanks smoser with the path tweaked to reference the mount-image-callback script thanks. | 17:11 |
blackboxsw | have a quick errand, back in 20 | 17:22 |
=== blackboxsw is now known as blackboxsw_bbl | ||
=== sambetts is now known as sambetts|afk | ||
smoser | k | 17:29 |
ahasenack | blackboxsw_bbl: smoser reading the etherpad and backlog | 17:42 |
=== blackboxsw_bbl is now known as blackboxsw | ||
blackboxsw | meh, looks like my xenial flavor of lxc doesn't accept push -p | 17:51 |
ahasenack | smoser: ok, read it all, +1 on the plan to go down the list and note which bug we are verifying | 17:54 |
ahasenack | smoser: https://bugs.launchpad.net/cloud-init/+bug/1674946 has been commented by someone saying the fix worked | 17:54 |
smoser | blackboxsw, yeah. you can use the snap i think for lxd | 17:55 |
ahasenack | not the reporter, though | 17:55 |
smoser | but migth still have the issue with mount-image-callback there. | 17:55 |
smoser | ahasenack, so what i'd do is just comment: | 17:55 |
smoser | I'm marking this verification-done-xenial based on comment X | 17:56 |
smoser | and then do that. | 17:56 |
ahasenack | ok | 17:56 |
smoser | ahasenack, blackboxsw i think you are both aware, but you tag with 'verification-done-xenial' and 'verification-done-yakkety' rather than just 'verification-done' | 18:03 |
ahasenack | right | 18:03 |
blackboxsw | smoser, hmm looks like my xenial proposed validation fell over the for https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/1674766 http://pastebin.ubuntu.com/24375361/ | 18:04 |
blackboxsw | the pastbin indicates ds-identify didn't process BigStep support right? | 18:04 |
blackboxsw | I'll double check that I have setup the seed files properl | 18:05 |
smoser | blackboxsw, right. let me try too | 18:05 |
blackboxsw | smoser this might be because I had started this lxc in the first place before adding the seed files in an attempt to work around my lxc file push not supporting -p option | 18:06 |
smoser | quite possibly. yes. | 18:07 |
blackboxsw | start would have run through cloud-init I expect | 18:07 |
blackboxsw | yeah sorting my lxc command version | 18:07 |
ahasenack | it's good to see the "before" case | 18:10 |
smoser | blackboxsw, yeah. so when working with cloud-init and lxc i have some tools (like the lxc-propsed-snapshot) that help you to adjust a container without starting it. | 18:18 |
smoser | the other option is to start it and then "clean" it. | 18:18 |
smoser | the bin/do-reboot script there helps do the clean | 18:22 |
ahasenack | smoser: config-drive newbie here, sorry | 18:23 |
ahasenack | how do I pass that to a container? | 18:23 |
ahasenack | a config-drive configuration, and where can i find out what it should look like? | 18:23 |
blackboxsw | TIL, thx smoser | 18:24 |
ahasenack | going through https://bugs.launchpad.net/nova-lxd/+bug/1673411 for some reference, looks like it has a helper script attached | 18:25 |
ahasenack | ah, it's in tools/ | 18:25 |
* ahasenack sees the commented bits about network_data.json | 18:30 | |
smoser | i've been walking up from the bottom | 18:34 |
smoser | but i'm going to leave https://bugs.launchpad.net/cloud-init/+bug/1665694 for one of you as a introduction to integration test suite | 18:34 |
blackboxsw | smoser, great. ok finally sorted things and have a zesty box at my disposal | 18:57 |
blackboxsw | lxc file pull $name/run/cloud-init/result.json - | 18:57 |
blackboxsw | { | 18:57 |
blackboxsw | "v1": { | 18:57 |
blackboxsw | "datasource": "DataSourceBigstep", | 18:57 |
blackboxsw | "errors": [] | 18:57 |
blackboxsw | } | 18:57 |
blackboxsw | } | 18:57 |
blackboxsw | ok progress. now time to do some work | 18:58 |
smoser | woot! | 18:59 |
smoser | blackboxsw, you can take the https://bugs.launchpad.net/cloud-init/+bug/1665694 | 18:59 |
blackboxsw | will do thx | 18:59 |
ahasenack | smoser: can I trigger this bug with just lxd, faking hyperv somehow? https://bugs.launchpad.net/cloud-init/+bug/1674946 | 19:08 |
ahasenack | I have a config drive in /config-drive in this lxd, with network_data.json, but it looks like it's not being read (I want to trigger the bug first, then check the fix) | 19:09 |
ahasenack | or maybe I'm being hit by https://bugs.launchpad.net/nova-lxd/+bug/1673411, but this cloud-init in -proposed should have that fix | 19:12 |
smoser | reading | 19:12 |
ahasenack | which is also in our list | 19:12 |
ahasenack | ah | 19:13 |
ahasenack | ok, got it | 19:13 |
smoser | ahasenack, (/me realizes for the first time that you have different internal and external personalities... andreas -> ahasenack) | 19:13 |
ahasenack | I wanted to test the "before" case, but I can't because I need the fixed cloud-init for it to parse /config-drive according to #1673411 | 19:13 |
ahasenack | smoser: :) | 19:13 |
ahasenack | andreas was already taken here, my ninjas didn't find him | 19:13 |
smoser | so before the new version, you should be able to populate /var/lib/cloud/data/seed/ConfigDrive | 19:14 |
smoser | i think | 19:14 |
smoser | (with the same contents as /config-drive) | 19:14 |
ahasenack | smoser: ok | 19:14 |
smoser | and it should look there. | 19:14 |
ahasenack | yay | 19:33 |
ahasenack | I haz result.json | 19:33 |
blackboxsw | interesting in running tox on the cloud-init tests is that my cd drive started spinning up. wondering if we're leaking some calls to mount that drive from the unit tests | 19:40 |
smoser | blackboxsw, it probably is. | 19:41 |
smoser | someone working on freebsd found one of those the other day ... let me se | 19:41 |
* ahasenack injecting errors to see what happens | 19:44 | |
ahasenack | "Unknown network_data link type: dvs-andreas-was-here", | 19:44 |
ahasenack | good | 19:45 |
smoser | nice ahasenack ! | 19:46 |
ahasenack | you made an MP against upstream to stop sending these deprecated link types, right | 19:47 |
ahasenack | https://review.openstack.org/#/c/400883/2 | 19:47 |
smoser | yeah, and it just made it into trunk | 19:47 |
smoser | they're really not "deprecated". as they're real and useful link types at the hypervisor level | 19:48 |
smoser | but not at the guest level | 19:48 |
ahasenack | right, wrong choice of words | 19:49 |
ahasenack | smoser: is cloud-init also responsible for creating the ubuntu user? | 19:50 |
smoser | yes | 19:50 |
smoser | > precise at least | 19:50 |
ahasenack | smoser: ok, because once I injected that failure in the network, the ubuntu user wasn't created: https://pastebin.canonical.com/185908/ | 19:50 |
smoser | yeah. if the datasource traces it wont go on | 19:51 |
ahasenack | ok | 19:51 |
smoser | well, in that way at least. | 19:51 |
ahasenack | sorry about the canonical pastebin | 19:51 |
ahasenack | habit | 19:52 |
ahasenack | and bookmarks | 19:52 |
smoser | ahasenack, well you can just use 'pastebinit' now! | 19:55 |
smoser | i am constantly doing: | 19:55 |
smoser | xsel -o | pastebinit | 19:55 |
ahasenack | indeed! | 19:55 |
ahasenack | finally | 19:55 |
blackboxsw | smoser, looks like I'm hitting a dependency error trying to install pylxd as the tox -e citest environment | 19:55 |
smoser | (highlight something and then paste and paste the url to something) | 19:55 |
smoser | hm.. | 19:56 |
blackboxsw | http://paste.ubuntu.com/24375992/ | 19:56 |
smoser | i suspect you need python3-dev | 19:56 |
rharper | blackboxsw: ping powersj ; he's worked through those before | 19:56 |
smoser | yeah | 19:56 |
blackboxsw | powersj, has been invoked :) | 19:56 |
smoser | tox is wonderful | 19:57 |
smoser | until you have c extentions | 19:57 |
smoser | then it becomes honestly kind of silly in my opinion | 19:57 |
smoser | but still better than lots of other stuff | 19:57 |
smoser | blackboxsw, install python3-dev with apt | 19:57 |
smoser | and try again | 19:57 |
smoser | pylxc needs c bindings to lxc and thus the sillynees. python3-dev is pretty light though. | 19:58 |
blackboxsw | getting closer. ok new traceback, missing C headers as well. I'll iterate over these to get them fixed. and report a list of the deps I had to install to get things working | 19:59 |
blackboxsw | ok needed only libssl-dev too. So, the list of packages I've installed on a fresh zesty system: ls | 20:04 |
blackboxsw | lxc1, cloud-utils, devscripts, python3-dev, libssl-dev | 20:05 |
nacc | we use lxc1? | 20:06 |
smoser | lxc1 provids lxc-usernsexec | 20:10 |
smoser | which is used by mount-image-callback lxd:foo | 20:10 |
smoser | to get the right user namespace when it chroots to the target | 20:10 |
rharper | nacc: yeah; I suggested we get lxd to implement lxc-usernsexec but I forget why they said not to | 20:10 |
smoser | blackboxsw, http://paste.ubuntu.com/24376111/ | 20:17 |
smoser | so inside an lxc container ^ worked. | 20:17 |
smoser | well, it shows me jumping into the container | 20:18 |
smoser | but that should list everything you need on xenial which i suspect is the same on zesty | 20:18 |
blackboxsw | excellent, will not distrupt my other systems then | 20:19 |
smoser | blackboxsw, well. you dont *really* want to run inside lxc | 20:20 |
smoser | because then you can't use zfs | 20:20 |
smoser | :) | 20:20 |
smoser | i'm not certain if it works or not fully. but lack of zfs is reason enough for me to not want to do it. | 20:20 |
blackboxsw | certainly. makes sense | 20:22 |
smoser | if you want... you can walk the merge proposal path ... | 20:22 |
smoser | with a doc enhancement to doc/rtd/topics/tests.rst | 20:22 |
smoser | with the above | 20:22 |
smoser | (the dependencies needed and such) | 20:23 |
blackboxsw | +1 smoser sure I'll put up an mp. | 20:23 |
ahasenack | smoser: ok, so I verified xenial and yakkety and added the respective verification-done-<release> tags. Someone else will flip the verification-needed tag into verification-done? | 20:30 |
ahasenack | ops, for bug https://bugs.launchpad.net/nova-lxd/+bug/1673411 that is | 20:30 |
smoser | ahasenack, no. you drop the verification-needed. | 20:35 |
smoser | changing it to verification-done-<x> | 20:35 |
smoser | really with 2 srus in flight at the same time... there really should be verificaton-needed-xenial and verification-needed-yakkety | 20:35 |
smoser | the sru process really just doesn't handle the different releases at once all that well | 20:36 |
ahasenack | smoser: how do they know they should hold pushing this to updates since we have the other bugs for the same package still being verified? Via debian/changelog? | 20:37 |
smoser | ahasenack, https://people.canonical.com/~ubuntu-archive/pending-sru.html | 20:51 |
smoser | look for cloud-init there. after everything gets verified, all its bugs should be green or gold (i dont understand the difference) | 20:52 |
ahasenack | ok, I see the proposed package associated with all the bugs | 20:52 |
blackboxsw | smoser, so I added verification-done-xenial verfication-done-yakkety tags to https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/1674766 | 20:56 |
Rafael | Hello folks | 21:00 |
=== Rafael is now known as Guest678 | ||
Guest678 | can somebody help to clarify a doubt that I have | 21:00 |
Guest678 | do I need any special configuration to enable cloud-init to reset SSH keys? | 21:01 |
Guest678 | The first time I create an instance it assigns the configure SSH key to it, however, if I change the key, cloud-init is not updating it in the VM | 21:01 |
smoser | change the key ? | 21:02 |
smoser | hi Raboo | 21:02 |
smoser | bah. sorry Raboo tab complete fail. | 21:02 |
Guest678 | yeap | 21:04 |
Guest678 | I mean change the public key that is inject into the VM | 21:04 |
Guest678 | When the VM is created cloud-init succesfuly install the keys under ~/.ssh/authrozied_keys | 21:05 |
Guest678 | However, if I change the keys in my cloud orchestrator, and then reboot the VM, cloud-init is not changing the keys | 21:05 |
smoser | Guest678, yeah, that is only done on first instance boot. | 21:06 |
Guest678 | I am guessing it has something to do with these logs "" | 21:06 |
Guest678 | yep | 21:06 |
Guest678 | that is what I thought from reading the log files | 21:06 |
smoser | it is not kept in sync. that is a feature that we'd like to have at some point. | 21:06 |
Guest678 | yes there a way to configure this? | 21:06 |
Guest678 | ah | 21:06 |
smoser | um... maybe. | 21:06 |
Guest678 | is there a work around? | 21:07 |
Guest678 | I thought that there would be a way to indicate which modules I want to run once per instance, or once per boot, or constantly with the VM running | 21:07 |
smoser | actually, not easily. you could turn the ssh config module (which is what puts those keys into the default user's .ssh/authorized_keys) | 21:08 |
smoser | but changing that to run PER_ALWAYS (boot) would be not ideal | 21:08 |
smoser | as that is the same thing that wipes the host keys | 21:08 |
smoser | which you probably do not want ot happen every boot | 21:08 |
smoser | what cloud provider is this ? | 21:09 |
Guest678 | ah | 21:09 |
Guest678 | the same script that configures SSH keys is setting up host keys? | 21:09 |
Guest678 | CloudStack | 21:09 |
Guest678 | the module you are talking about is the one called "ssh"? | 21:09 |
Guest678 | I noticed some other module called "ssh-authkey-fingerprints" | 21:10 |
smoser | blackboxsw, on https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/1674766 | 21:10 |
smoser | (and the others you have done) | 21:11 |
smoser | the sru person will want to see a comment at least stating "i verified both xenial and yakkety following the test case template above" | 21:11 |
smoser | i like to copy and paste evidence of having done that | 21:11 |
smoser | 21:11 | |
smoser | https://bugs.launchpad.net/cloud-init/+bug/1674685https://bugs.launchpad.net/cloud-init/+bug/1674685 | 21:11 |
smoser | Guest678, yeah, that does sound silly doesnt it :) | 21:12 |
smoser | yes, its ssh. we really should separate that out so you could tweak one wihtout the other. | 21:12 |
smoser | but as it is right now... the datasources cache the data on disk (so as to not rely upon a metadata service on reboot) | 21:12 |
Guest678 | hmm | 21:12 |
smoser | so.. they'd just re-aply the cached data. | 21:12 |
smoser | this is a feature we'd like to have, but its not present now. sorry . | 21:13 |
blackboxsw | +1 smoser will add that now and close those bugs appropriately | 21:13 |
smoser | well, you dont change the state of the bugs.. | 21:13 |
blackboxsw | s/close/tag | 21:13 |
Guest678 | thanks @smoser | 21:13 |
smoser | once the bug moves from -prpoosed to -updates it will automatically do that. | 21:13 |
smoser | ahasenack, https://bugs.launchpad.net/cloud-init/+bug/1570325 | 21:17 |
smoser | you can do that one... that is a integration test also. (for learning to at least run the integration test) | 21:17 |
ahasenack | deal | 21:17 |
smoser | ok. /me has to run. | 21:21 |
ahasenack | ok, https://bugs.launchpad.net/cloud-init/+bug/1674946 verified | 21:27 |
ahasenack | this will go much faster from now on | 21:27 |
ahasenack | ok, I'm EOD | 21:31 |
* blackboxsw is spending some time walking through those integration tests on cloud-init | 21:45 | |
blackboxsw | see you smoser/ahasenack | 21:45 |
powersj | blackboxsw: https://cloudinit.readthedocs.io/en/latest/topics/tests.html also worth a read | 21:46 |
blackboxsw | excellent | 21:50 |
blackboxsw | keep 'em coming | 21:50 |
blackboxsw | actually, that's what I'm going to put up a simple doc mp for | 21:51 |
blackboxsw | just trying to wrap up some verification first | 21:51 |
powersj | blackboxsw: are you familiar with tox and running it? | 21:52 |
powersj | both cloud-init and curtin use tox as a wrapper around our unit tests and style tests. When we get a MR that is what we expect to pass | 21:52 |
blackboxsw | just a bit, TIL about environments. you guys have a bit more tox.ini config options/setup than services I've worked on | 21:52 |
powersj | you can look at tox.ini in the root dir of both projects | 21:52 |
powersj | ah ok :) | 21:52 |
dpb1 | powersj: MR... wth | 21:53 |
powersj | O.o | 21:53 |
powersj | merge request? | 21:53 |
dpb1 | step child of MP and PR | 21:53 |
powersj | lol | 21:53 |
blackboxsw | haha | 21:53 |
* powersj is split brain due to github and launchpad usage | 21:54 | |
dpb1 | :) | 21:54 |
=== rangerpb is now known as rangerpbzzzz | ||
powersj | Looks like pylint 1.7 was released today O.o | 23:54 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!