| *** Jeffrey4l has left #kata-dev | 00:20 | |
| *** jpeach has quit IRC | 03:55 | |
| *** pcaruana has joined #kata-dev | 04:56 | |
| *** sameo has joined #kata-dev | 05:40 | |
| *** igordc has quit IRC | 06:32 | |
| *** sameo has quit IRC | 06:55 | |
| *** tmhoang has joined #kata-dev | 07:22 | |
| *** jodh has joined #kata-dev | 07:39 | |
| *** stefanha has quit IRC | 07:44 | |
| *** stefanha has joined #kata-dev | 07:51 | |
| *** gwhaley has joined #kata-dev | 07:56 | |
| *** davidgiluk has joined #kata-dev | 08:00 | |
| *** tmhoang has quit IRC | 08:01 | |
| *** tmhoang has joined #kata-dev | 08:02 | |
| *** sameo has joined #kata-dev | 08:05 | |
| *** davidgiluk has quit IRC | 08:23 | |
| *** davidgiluk has joined #kata-dev | 08:25 | |
| kata-irc-bot | <graham.whaley> morning @salvador.fuentes @eric.ernst - I see on a recent tests repo PR we have a new CI item 'ci/dockercloud' (see https://github.com/kata-containers/tests/pull/1725). Any info on that? It looks like its hotlink is not functional yet - when I follow the 'Details' link, it has a large magic URL, but ends up being redirect/resolved to the top level docker hub :( | 08:29 |
|---|---|---|
| gwhaley | morning davidgiluk. just to note, somebody over on the kata-general channel was asking if there were performance numbers for virtio-fs vs 9p etc. iirc, there were some posted with the original announcement. I had a look on the website but did not find any. If you happen to have a link to any then I'll post it over on the other channel. thx! | 08:32 |
| davidgiluk | gwhaley: I *think* that's the latest set we released | 08:32 |
| gwhaley | don't suppose you know when that was davidgiluk? - then I can have a peek in the kata email archive (I think that's where it was posted?) | 08:33 |
| davidgiluk | gwhaley: I'd have to do the same, but let me just peak in the virito-fs archives first | 08:34 |
| *** dklyle has quit IRC | 08:35 | |
| *** david-lyle has joined #kata-dev | 08:35 | |
| davidgiluk | gwhaley: https://lwn.net/ml/linux-fsdevel/20181210171318.16998-1-vgoyal@redhat.com/ is the original kernel patch release | 08:35 |
| davidgiluk | gwhaley: The v2 doesn't have the numbers on, so I think that is the latest posted | 08:36 |
| gwhaley | great, thx! | 08:37 |
| *** gwhaley has quit IRC | 10:58 | |
| *** lpetrut has joined #kata-dev | 12:05 | |
| *** gwhaley has joined #kata-dev | 12:24 | |
| *** gwhaley has quit IRC | 12:24 | |
| *** gwhaley has joined #kata-dev | 12:28 | |
| *** lpetrut has quit IRC | 12:39 | |
| *** pcaruana has quit IRC | 13:37 | |
| kata-irc-bot | <salvador.fuentes> @graham.whaley no idea of that new CI item :O maybe @eric.ernst has info | 13:42 |
| *** pcaruana has joined #kata-dev | 13:54 | |
| *** sameo has quit IRC | 14:14 | |
| *** sameo_ has joined #kata-dev | 14:14 | |
| *** devimc has joined #kata-dev | 14:18 | |
| kata-irc-bot | <graham.whaley> @salvador.fuentes @gmmaharaj @eric.ernst @james.o.hunt - I have put together the framework of a Jenkinsfile as an RFC around how we might put this together for Kata. I worked out how to grab the labels off github and allow a 'fast path' using a label (such as, if you set a label 'skip-ci' on the PR, the CI will short circuit return). I can run it locally, but can't run it on the test server as I think having a space in the name of | 14:49 |
| kata-irc-bot | the github token credential is upsetting it ;). The file is at https://github.com/grahamwhaley/kata-containers-runtime/blob/20190604_jenpipe/Jenkinsfile - can you have a peek and let me know what you think as an initial idea? I can always PR this for review to the main repo (we might want to store it in the CI repo if we use it for all repos...). I have more ideas to make it more 'flexible' to be multi-distro, multi-repo friendly as well, so | 14:49 |
| kata-irc-bot | maybe we can end up with one (large) Jenkinsfile to rule them all... | 14:49 |
| kata-irc-bot | <graham.whaley> @salvador.fuentes - I'll have a stab at working around the 'space in the name' issue, but if I can't work it out, I'll request we rename the token I guess ;) | 14:49 |
| kata-irc-bot | <james.o.hunt> @graham.whaley - ta. Can you point us at the best doc/site for grokking these pipelines? :slightly_smiling_face: | 14:51 |
| kata-irc-bot | <graham.whaley> @james.o.hunt heh. hmm, well, I can point you at the pipe docs, and then some of it is 'groovy'... generally, no, I'm not sure there is an easy entree to pipelines... I tried to comment it heavily to help.... https://jenkins.io/doc/book/pipeline/ | 14:52 |
| *** jodh has quit IRC | 14:59 | |
| *** sameo_ has quit IRC | 15:51 | |
| *** pcaruana has quit IRC | 16:01 | |
| *** igordc has joined #kata-dev | 16:03 | |
| gmmaha | davidgiluk: thanks for the tip on the hotplug. we now have a patch and hotplug works well with virtio-fs | 16:18 |
| gmmaha | in the tiny little test i did with it. | 16:18 |
| davidgiluk | gmmaha: Treat it with caution; if you look at the code you'll see the bits that aren't thread safe and it'll stop working :-) | 16:19 |
| gmmaha | davidgiluk: hehe. i need to look at the code.. but after 3 weeks. now all i can think of is hot and humid weather and beaches. | 16:21 |
| gmmaha | but those would be good to sort out.. very soon someone is going to run this in production and trying it out! ;) | 16:21 |
| davidgiluk | gmmaha: Yeh I'll look at that - I know how to solve most of it I think | 16:29 |
| gmmaha | davidgiluk: ohh that is fantastic. | 16:29 |
| kata-irc-bot | <salvador.fuentes> hi @graham.whaley, just saw the pipeline, it looks good, although we would like to run static-checks always, right? I see that you also skip them when `SKIPTESTS=true`, wdyt? | 16:33 |
| kata-irc-bot | <graham.whaley> fuentess: yeah, it is just a PoC right now - I suspect we do always want to run the static checks, yets :slightly_smiling_face: | 16:34 |
| kata-irc-bot | <graham.whaley> let me PR this tomorrow for some RFC, as I have some further questions we need to discuss about how we might make a table/matrix of which distros run which tests for which repos - and we can probably do that in a number of ways... | 16:34 |
| kata-irc-bot | <graham.whaley> - open code it in groovy/pipeline | 16:35 |
| kata-irc-bot | <graham.whaley> - store the data in a groovy map, and parse it at runtime | 16:35 |
| kata-irc-bot | <graham.whaley> - store the data in a yaml, json or csv file and parse it from there | 16:35 |
| kata-irc-bot | <graham.whaley> I'll try to write it up tomorrow so we can discuss :slightly_smiling_face: | 16:35 |
| kata-irc-bot | <graham.whaley> I'm just trying to see if I can get the list of changed files dynamically (without having to resort to `git` yet), so we can try a 'docs only' smart check as well. | 16:36 |
| gmmaha | brtknr: this is what i am seeing as performance between 9p and virtio-fs in my small test | 16:37 |
| gmmaha | http://paste.openstack.org/show/753148/ | 16:37 |
| davidgiluk | gmmaha: IMHO that test is way way too simple to be meaningful; even if it is very nice for us | 16:39 |
| kata-irc-bot | Action: graham.whaley mumbles at `dd` as a benchmark ;) You could always try | 16:39 |
| kata-irc-bot | <graham.whaley> wait for it... | 16:39 |
| kata-irc-bot | <graham.whaley> https://github.com/kata-containers/tests/tree/master/metrics/report which does include a fairly simple `fio` based check | 16:40 |
| kata-irc-bot | <graham.whaley> ;) | 16:40 |
| kata-irc-bot | <graham.whaley> I know @gmmaharaj, you are trying to get on a plane - not expecting you to be running this right now! | 16:40 |
| kata-irc-bot | <salvador.fuentes> @graham.whaley cool, thank you | 16:41 |
| kata-irc-bot | <gmmaharaj> ohh i got 18 hours. i can test it | 16:41 |
| kata-irc-bot | <graham.whaley> arf - do it in parallel to something else - the fio tests take a while to run (such is their nature) | 16:41 |
| kata-irc-bot | <graham.whaley> if you've not used the report generator before - the grabdata.sh takes args to restrict what tests it runs - so there is something like `-s` to just run storage tests.. (check before trusting that) | 16:42 |
| kata-irc-bot | <gmmaharaj> @graham.whaley so build the docker image, change the entrypoint, run the container with virtio-fs and get numbers. | 16:44 |
| kata-irc-bot | <gmmaharaj> right? | 16:44 |
| *** pcaruana has joined #kata-dev | 16:45 | |
| kata-irc-bot | <graham.whaley> gmmaha: set up your docker how you want it - set the RUNTIME, run `grabdata.sh -s`, move the results to a nicely named subdir.... repeat for 2nd docker setup (and 3rd, 4th etc.)..... when you have all the data, run `makereport.sh` to get a pdf of analysis and comparison. The README explains, and the script tells you what to do when you run it. | 16:46 |
| kata-irc-bot | <graham.whaley> so, no need to build any docker images (unless you are testing changes in a docker image, in which case you would need to make sure the local image is the same name as the tests expect, or edit the tests). The generator is for testing against different docker setups (normally say different runtime verisons etc.) - so you set up your system as you'd like it tested, and run the tests.. | 16:47 |
| *** igordc has quit IRC | 16:49 | |
| kata-irc-bot | <gmmaharaj> thanks @graham.whaley going to test fio between 9p and virtiofs | 16:59 |
| gmmaha | davidgiluk: true.. but the laziest option it was.. ;) | 17:00 |
| *** altlogbot_2 has quit IRC | 17:00 | |
| *** altlogbot_2 has joined #kata-dev | 17:00 | |
| davidgiluk | gmmaha: Oh yes, feel free to say how we're great and fast :-) | 17:01 |
| *** gwhaley has quit IRC | 17:01 | |
| gmmaha | :D wihout hesitation.. ;) | 17:02 |
| *** irclogbot_1 has quit IRC | 17:02 | |
| *** irclogbot_0 has joined #kata-dev | 17:02 | |
| *** tmhoang has quit IRC | 17:05 | |
| kata-irc-bot | <gmmaharaj> @graham.whaley just went through the pipeline. it looks pretty good.. also once you have tested the package on the primary repo, you can save the kata-tarball and pass it along to the next stage and all the distros, so that you don't have to re-build them all and just setup the env.. It saves time, but we lose the functionality of testing compile on the distros. it can be something we can think about on a later stage. | 17:16 |
| kata-irc-bot | <gmmaharaj> also not sure if there is a way to interrupt a running CI job to stop it if the patch has a newer version uploaded.. i know we do it today.. Just wondering how that works with pipeline. | 17:17 |
| *** igordc has joined #kata-dev | 17:35 | |
| *** devimc has quit IRC | 17:48 | |
| kata-irc-bot | <gmmaharaj> ls | 18:11 |
| *** pcaruana has quit IRC | 18:45 | |
| *** pcaruana has joined #kata-dev | 19:05 | |
| *** davidgiluk has quit IRC | 19:06 | |
| *** sameo_ has joined #kata-dev | 19:50 | |
| *** sameo_ has quit IRC | 20:38 | |
| brtknr | gmmaha: looks good! i got something similar on a local path on the hypervisor presented to a kata container... have you tried a network disk mounted on the hypervisor and tried presenting that to kata? thats where things seem to get interesting for me in the case of virtiofs | 21:38 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!