| dmsimard | boxrick1: ideally everything would be the same version, I think you're bound to run into problems if you run a callback somewhere that doesn't have the latest sql migrations/models | 00:05 |
|---|---|---|
| *** openstackgerrit has quit IRC | 11:33 | |
| *** themurph has quit IRC | 13:31 | |
| *** themurph has joined #ara | 13:34 | |
| *** fignew has joined #ara | 13:46 | |
| dmsimard | boxrick1: did you give the patch a try ? | 13:50 |
| boxrick1 | Yea, it feels faster. But when I timed it with a big page it wasn't noticeably so. But with with a page size of around 25 its now usable. | 13:52 |
| boxrick1 | So id say a decent improvement overall, just not with massive pages. | 13:52 |
| boxrick1 | I stuck some UI suggestions on the Openstack story page. | 13:53 |
| dmsimard | boxrick1: yeah, I saw | 13:53 |
| dmsimard | boxrick1: there's more (and more involved) performance optimization opportunities in the future | 13:54 |
| dmsimard | I need to write them down, saw a few yesterday when I worked on that patch | 13:54 |
| boxrick1 | Well I think a dynamic loading when you click, as opposed to at page load would help considerably | 13:55 |
| dmsimard | boxrick1: for the home page ? | 13:56 |
| boxrick1 | On the reports page | 13:56 |
| boxrick1 | The home page is not something I really use, its just a big title with loads of stats. | 13:56 |
| dmsimard | boxrick1: yeah. | 13:57 |
| dmsimard | There is only so much I can do for the reports page, I am by no means a professional "full stack" developer but would definitely welcome experts to help the application scale better | 13:57 |
| boxrick1 | Ah I appreciate that, these are hopefully just little things to make a big improvement. Your web skills seem decent enough to me. | 13:58 |
| dmsimard | There's probably a bunch of awesome ajax and react and nodejs stuff that could be done to make things snappy and all but it's really not my cup of tea :p | 13:59 |
| boxrick1 | Meh, keep the tool functional and nail the basics. I would have personally kept things along the lines of page load per play even to keep it simp.e | 14:00 |
| boxrick1 | As you say leave the flashy GUIs to someone who does it as a day job. | 14:00 |
| dmsimard | The whole "centralized" or "persistent" ARA story needs attention and I'm probably first guilty of that, I want to dogfood it but haven't got around to it | 14:02 |
| boxrick1 | You mean in the fashion I use it? | 14:03 |
| dmsimard | I've been using it in mostly ephemeral and CI context where things definitely don't scale to hundreds of playbooks and millions of records | 14:03 |
| *** tbielawa has joined #ara | 14:51 | |
| *** openstackgerrit has joined #ara | 15:05 | |
| openstackgerrit | David Moreau Simard proposed openstack/ara master: Improve count performance https://review.openstack.org/450449 | 15:05 |
| openstackgerrit | Merged openstack/ara master: Improve count performance https://review.openstack.org/450449 | 15:30 |
| *** _dev has joined #ara | 15:51 | |
| *** tbielawa has quit IRC | 16:54 | |
| *** tbielawa has joined #ara | 16:57 | |
| *** tbielawa is now known as tbielawa|lunch | 18:13 | |
| dmsimard | mnaser, logan-: huh, I just reproduced the http/https thing completely by accident | 19:03 |
| mnaser | dmsimard oh cool, have a public link? : | 19:04 |
| mnaser | :p | 19:04 |
| dmsimard | mnaser: https://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13/reports/index.html | 19:04 |
| mnaser | Mixed Content: The page at 'https://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13/reports/index.html' was loaded over HTTPS, but requested an insecure resource 'http://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13/file/78a14335-aee2-4c0c-902c-317b7b324fc7/'. This request has been blocked; the content must be served over HTTPS.o | 19:05 |
| dmsimard | It's not from jenkins htmlpublisher and it's not hosted as a jenkins artifact, it's statically generated and rsync'd to a dumb server | 19:05 |
| mnaser | ok interesting | 19:05 |
| mnaser | could it be possible iframe defaults to http | 19:05 |
| dmsimard | have no clue | 19:06 |
| dmsimard | trying to see | 19:06 |
| dmsimard | the iframe uses a relative link | 19:06 |
| mnaser | yeah i saw that too | 19:06 |
| dmsimard | i.e, <iframe id="result_iframe" class="iframe" data-src="../result/index.html" src="../result/bd9b47ba-4317-4e08-b5bf-45f89f2e3576"></iframe> | 19:06 |
| mnaser | and for some reason | 19:07 |
| dmsimard | so I'm not sure how that can end up happening | 19:07 |
| mnaser | the https request gets cancelled | 19:07 |
| mnaser | and then does the http one which gets rejected | 19:07 |
| dmsimard | googling this is the plague, everyone wants to include an http iframe in a https site | 19:07 |
| mnaser | this is so strange | 19:09 |
| mnaser | dmsimard if you click the network tab and hit smoething to open a modal | 19:09 |
| mnaser | it goes from (canceled) to (blocked:mixed-content) | 19:09 |
| mnaser | making 2 requests, an http and https | 19:09 |
| dmsimard | mnaser: I don't have access to the server it's hosted on, maybe there's something going on with the webserver config | 19:10 |
| dmsimard | I know it's nginx though | 19:10 |
| mnaser | hey hey hey | 19:10 |
| mnaser | i found it | 19:10 |
| mnaser | :D | 19:10 |
| dmsimard | show me | 19:10 |
| mnaser | ok dmsimard are you ready | 19:10 |
| dmsimard | I am | 19:10 |
| dmsimard | let me guess | 19:10 |
| dmsimard | a typo | 19:10 |
| mnaser | curl -vvv https://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13/file/78a14335-aee2-4c0c-902c-317b7b324fc7 | 19:10 |
| mnaser | <p>The document has moved <a href="http://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13/file/78a14335-aee2-4c0c-902c-317b7b324fc7/">here</a>.</p> | 19:11 |
| mnaser | redirecting it to http | 19:11 |
| mnaser | the lack of trailing / | 19:11 |
| mnaser | results in a 301 | 19:11 |
| mnaser | which does not redirect to https | 19:11 |
| mnaser | (i would say this is a miconfigured server, but if you added a trailing /, you avoid this) | 19:11 |
| dmsimard | ehhhhhhhhhhhhhhhhhhhhhhhh | 19:11 |
| dmsimard | the server is not wrong | 19:11 |
| mnaser | yeah but its redirecting to http: not https: | 19:12 |
| dmsimard | exists: https://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13/file/78a14335-aee2-4c0c-902c-317b7b324fc7/index.html | 19:12 |
| dmsimard | doesn't exist: https://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13/file/78a14335-aee2-4c0c-902c-317b7b324fc7 | 19:12 |
| dmsimard | now, don't get me wrong, the web server behavior is definitely weird here | 19:12 |
| mnaser | no thats normal | 19:13 |
| mnaser | apache always does this | 19:13 |
| mnaser | GET /artifacts/rdo/poc-dmsimard-kolla-build/13/file/78a14335-aee2-4c0c-902c-317b7b324fc7 | 19:13 |
| mnaser | apache sees that there is no file with that name, but a folder exists with that name | 19:13 |
| mnaser | responds with a 301 to /artifacts/rdo/poc-dmsimard-kolla-build/13/file/78a14335-aee2-4c0c-902c-317b7b324fc7/ | 19:13 |
| mnaser | because there is an index.html inside of it, it responds with the index file | 19:13 |
| mnaser | (technically no trailing slash => file, trailing slash => folder) | 19:14 |
| dmsimard | mnaser: yeah | 19:14 |
| dmsimard | logan-: fyi we found your problem ^ | 19:14 |
| dmsimard | now to fix it ... | 19:15 |
| *** tbielawa|lunch is now known as tbielawa | 19:15 | |
| mnaser | https://github.com/openstack/ara/blob/master/ara/templates/report_list.html#L28 -- maybe replace those calls with url_for | 19:15 |
| mnaser | and get rid of the magic replacement logic | 19:16 |
| mnaser | (and you could even technically reduce it to a single modal too, because you're just dynamically loading the iframe with whatever data-load url contains | 19:16 |
| dmsimard | mnaser: can't get rid of url_for, that's the magic that makes flask-frozen work | 19:20 |
| mnaser | oh im saying add it there instead of getting rid of it | 19:20 |
| dmsimard | mnaser: https://github.com/openstack/ara/blob/master/ara/templates/report_list.html#L423 | 19:20 |
| dmsimard | return $(this).data('src').replace('index.html','').concat(load + '/'); seems to work | 19:21 |
| dmsimard | at least it doesn't break the existing behavior, just need to check if it'd fix the issue | 19:21 |
| * dmsimard tests | 19:21 | |
| mnaser | i was thinking https://github.com/openstack/ara/blob/master/ara/templates/report_list.html#L28 -- {{ playbook.file.id }} => {{ url_for('playbook.show_playbook', id=playbook.file.id) }}, | 19:22 |
| dmsimard | mnaser: I can add it there, but then I need to add 15 slashes instead of one | 19:22 |
| mnaser | true | 19:22 |
| dmsimard | trying to re-generate with the fix | 19:23 |
| dmsimard | see what happens | 19:23 |
| dmsimard | mnaser: wfm https://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13-patched/reports/index.html | 19:27 |
| dmsimard | what a stupid bug | 19:27 |
| mnaser | wee :D | 19:27 |
| dmsimard | stupidest bug in ara's history so far imo | 19:27 |
| mnaser | :p | 19:27 |
| mnaser | glad you got a reproducable case that was accessible | 19:27 |
| mnaser | this would have been impossible if it was done behind closed doors | 19:27 |
| dmsimard | yeah, I mean, reproducing the problem is crucial to figuring out what the root cause is .. | 19:28 |
| dmsimard | and I did it by accident \o/ | 19:28 |
| dmsimard | this is timely, I was one patch away from cutting release candidate for 0.12.4 | 19:28 |
| dmsimard | I want to get rid of the goddamn sqlalchemy log messages in the foreground | 19:28 |
| dmsimard | mnaser: fwiw you have an example of the per-tag kolla-build thing I told you about | 19:30 |
| dmsimard | nope, not the right fix, almost there | 19:34 |
| mnaser | another error dmsimard ? | 19:34 |
| dmsimard | mnaser: doesn't work for the action column in the task results because the task results adds an anchor to highlight the line number | 19:34 |
| dmsimard | so we're appending a / after # | 19:35 |
| mnaser | eeek | 19:35 |
| dmsimard | guess I'll have to add it in the url_for | 19:35 |
| mnaser | yeah that'll probably eliminate any weird stuff | 19:36 |
| mnaser | and the replace logic | 19:36 |
| openstackgerrit | Devon Mizelle proposed openstack/ara master: adding postgresql text type "isms" https://review.openstack.org/450948 | 19:43 |
| _dev | oh, that's me! | 19:44 |
| _dev | a little uncomfortable having my full name in it, but w/e. had to fight gerrit enough. | 19:44 |
| dmsimard | _dev: o/ | 19:53 |
| _dev | hi! | 19:53 |
| dmsimard | _dev: that looks great at first glance I'll have a better look later | 19:54 |
| dmsimard | thanks a lot for jumping through the hoops, you have my eternal gratitude :) | 19:55 |
| _dev | thank you for a great project! | 19:56 |
| _dev | let me know if i need to change anything, I hope the changes to run_tests.sh work decently. | 19:56 |
| _dev | i didnt want to enforce use of docker to spin up pgsql, so its kind of left as an exercise to the user/jenkins. | 19:56 |
| _dev | but i documented how to do it in docker, at least. | 19:56 |
| dmsimard | _dev: yeah right now we're integration testing sqlite only but I want to do integration testing on mysql and pgsql as well | 19:57 |
| dmsimard | _dev: so just to show you how things work | 19:58 |
| dmsimard | in your patch, if you click on, say, "gate-ara-integration-latest-ubuntu-xenial" | 19:58 |
| dmsimard | that sends you to the job logs -- console.html is the job console (where i.e, run_tests.sh sends it's output) but we also collect various logs | 19:58 |
| dmsimard | http://logs.openstack.org/48/450948/1/check/gate-ara-integration-latest-ubuntu-xenial/149e3d9/ | 19:58 |
| dmsimard | if you go in "build" you can even see the statically generated report for whatever run_tests.sh has run | 19:59 |
| dmsimard | unit tests are provided by gate-ara-python27-ubuntu-xenial and pep8/bandit tests are provided by gate-ara-pep8-ubuntu-xenial | 20:00 |
| openstackgerrit | David Moreau Simard proposed openstack/ara master: Append slash to modal iframe URLs https://review.openstack.org/450965 | 20:08 |
| dmsimard | mnaser: hopefully that does it ^ I'll try it out now. | 20:09 |
| logan- | dmsimard: awesome stuff. Just reading the backlog | 20:09 |
| dmsimard | logan-: that was a weird problem | 20:10 |
| dmsimard | (╯°□°）╯︵ ┻━┻ | 20:10 |
| openstackgerrit | David Moreau Simard proposed openstack/ara master: Append slash to modal iframe URLs https://review.openstack.org/450965 | 20:13 |
| dmsimard | lgtm | 20:14 |
| dmsimard | logan-: fyi, reproduced here: https://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13/reports/index.html | 20:15 |
| dmsimard | and then with the patch ^: https://ci.centos.org/artifacts/rdo/poc-dmsimard-kolla-build/13-patched/reports/index.html | 20:15 |
| _dev | ah okay, so you all gather the logs up and bundle them with the review | 20:15 |
| _dev | thats pretty nifty. | 20:15 |
| dmsimard | _dev: after you've jumped through the hoops once and get used to the workflow it's pretty awesome :) | 20:16 |
| logan- | So it was more about fronting ARA with nginx than anything on the Jenkins side.. heh | 20:27 |
| dmsimard | logan-: you happen to run nginx too ? | 20:27 |
| logan- | Yeah my Jenkins is behind a nginx proxy | 20:28 |
| dmsimard | well there ya go | 20:28 |
| dmsimard | logan-: it'll land in 0.12.4 sometime this week | 20:28 |
| logan- | I suspect that's why we were seeing that same behavior | 20:28 |
| logan- | Cool | 20:28 |
| dmsimard | larsks: do you have any remote idea why we put that number there ? I have no recollection https://github.com/openstack/ara/blob/master/ara/db/versions/001_initial_revision.py#L80 | 20:29 |
| logan- | There's some other content security protection stuff that needs to be adjusted in Jenkins, I'll work on getting that documented after I'm able to get things working on the new version :) | 20:29 |
| *** tbielawa has quit IRC | 20:29 | |
| * larsks looks | 20:29 | |
| dmsimard | larsks: it's carried even from the original pull request https://github.com/dmsimard/ara/pull/89/commits/3680ea6ec1f712f38373ebd8c9f8a681a8b8898b | 20:30 |
| dmsimard | but the number looks too specific | 20:30 |
| dmsimard | it's probably something like sqlite vs mysql implementation of text() | 20:30 |
| larsks | dmsimard: that is the max value for an unsigned mysql MEDIUMINT | 20:31 |
| larsks | ...which is probably the max size of a text field. | 20:31 |
| larsks | I'll bet that's it. | 20:32 |
| dmsimard | larsks: something like sqlite was interpreting text() too small ? | 20:32 |
| dmsimard | larsks: so there's https://review.openstack.org/#/c/450948/1/ara/db/versions/001_initial_revision.py@80 | 20:32 |
| dmsimard | figure there'd be a better way around it | 20:32 |
| larsks | Is your suggestion there that maybe a with_variant could be used for mysql? | 20:34 |
| dmsimard | larsks: not familiar with how to properly handle different variants, I'm poking around olso.db and other projects to check if there's anything in particular | 20:36 |
| dmsimard | larsks: http://codesearch.openstack.org/?q=with_variant&i=nope&files=&repos= | 20:38 |
| dmsimard | so wrap the class | 20:39 |
| larsks | dmsimard: so, this is where it came from in ara: https://github.com/openstack/ara/commit/2850c1cd4746a81e6cb8023d17702d8da92402ed | 20:40 |
| dmsimard | ah nice find | 20:40 |
| larsks | git log -G 16777215 | 20:40 |
| dmsimard | well actually not quite | 20:41 |
| larsks | But yeah, the with_variant stuff seems like it might be cleaner. | 20:41 |
| dmsimard | that's for the result table, this is the host facts table | 20:41 |
| dmsimard | but same thing | 20:41 |
| larsks | Right. | 20:41 |
| dmsimard | oh I know who I can ask | 20:42 |
| dmsimard | harlowja: hai | 20:42 |
| harlowja | yo | 20:42 |
| dmsimard | harlowja: what'd be the proper way to handle a 3 way variance between sqlite/pgsql/mysql in migrations ? | 20:42 |
| dmsimard | harlowja: from looking at http://codesearch.openstack.org/?q=with_variant&i=nope&files=&repos= it looks like it's just one variant, not two | 20:42 |
| * harlowja redirects to zzzeek :-P | 20:42 | |
| harlowja | jump in #openstack-oslo ask zzzeek | 20:43 |
| * dmsimard jumps | 20:43 | |
| harlowja | he sorta created sqlalchemy | 20:43 |
| harlowja | so he should know | 20:43 |
| harlowja | lol | 20:43 |
| dmsimard | oh snap | 20:43 |
| dmsimard | harlowja: we have so many smart people involved in OpenStack it's great | 20:47 |
| harlowja | :-P | 20:47 |
| dmsimard | harlowja: there's the pypi guy too | 20:47 |
| dmsimard | was surpsied | 20:47 |
| harlowja | is he still around? | 20:47 |
| harlowja | though rackspace let him go :-/ | 20:47 |
| dmsimard | he got a new job somewhere | 20:47 |
| harlowja | *thought | 20:47 |
| harlowja | kk | 20:47 |
| harlowja | good | 20:47 |
| dmsimard | he was at HPE I think | 20:47 |
| harlowja | ya | 20:47 |
| dmsimard | harlowja: oh right he's at amazon now. | 20:48 |
| harlowja | interesting | 20:48 |
| _dev | seems weird, but maybe you can chain the with_variant? | 21:38 |
| _dev | like .with_variant(mysql.VARCHAR(), 'mysql').with_variant(somethingother, 'postgresql') | 21:39 |
| openstackgerrit | Merged openstack/ara master: Append slash to modal iframe URLs https://review.openstack.org/450965 | 22:18 |
| _dev | dmsimard: I just saw your comments on my review. I'm about to leave the office, so I'll push a commit or two tomorrow. | 22:27 |
| dmsimard | _dev: yeah no rush. | 22:28 |
| harlowja | hmmm, dmsimard when running with sqlite and now with the ability to sigterm running ansible, seem to be getting http://imgur.com/a/DvqWG after i did this :-P | 22:39 |
| harlowja | doesn't seem to kill ara, just throws that message up | 22:39 |
| harlowja | lol | 22:39 |
| dmsimard | harlowja: https://github.com/openstack/ara/commit/a44fc4e53abb39635841a3c997d54d991c5e343c | 22:40 |
| dmsimard | 0.12.4 | 22:40 |
| dmsimard | Soon (tm) | 22:40 |
| harlowja | ah | 22:40 |
| harlowja | k | 22:40 |
| harlowja | coool | 22:40 |
| harlowja | i think the loading task result page also spins? | 22:40 |
| harlowja | vs erroring out? | 22:40 |
| dmsimard | Yeah it'll spin endlessly | 22:41 |
| harlowja | kk | 22:41 |
| dmsimard | Haven't checked how to make it do something else | 22:41 |
| harlowja | np | 22:41 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!