[01:22:18] 3Wikimedia / 3Quality Assurance: No data received by Saucelabs from beta labs - 10https://bugzilla.wikimedia.org/68083 (10Juliusz Gonera) 3NEW p:3Unprio s:3normal a:3None Tests often fail because pages are not loading. Seems to be happening often on the login page. See: https://integration.wikimedia... [01:28:34] 3Wikimedia / 3Quality Assurance: Beta labs API not accessible from Saucelabs - 10https://bugzilla.wikimedia.org/68084 (10Juliusz Gonera) 3NEW p:3Unprio s:3normal a:3None The code that creates pages using API fails with: getaddrinfo: Name or service not known (Faraday::ConnectionFailed) It seems som... [03:55:32] 3Wikimedia / 3Quality Assurance: No data received by Saucelabs from beta labs - 10https://bugzilla.wikimedia.org/68083 (10Greg Grossmeier) p:5Unprio>3High [03:55:32] 3Wikimedia / 3Quality Assurance: Beta labs API not accessible from Saucelabs - 10https://bugzilla.wikimedia.org/68084 (10Greg Grossmeier) p:5Unprio>3High [07:01:30] zeljkof:Hi [07:01:41] Jagori: ready? [07:02:00] yup:) [07:02:31] Jagori: can you join the hangout? [07:02:45] yes [07:03:00] Jagori: I am already there [07:03:17] joining [07:08:57] Jagori: https://github.com/styleguide/ruby [07:37:38] Jagori: https://help.github.com/articles/ignoring-files#global-gitignore [07:42:43] Jagori: https://github.com/zeljkofilipin/dotfiles [08:02:48] Jagori: http://lists.wikimedia.org/pipermail/wikitech-l/2013-May/069672.html [10:48:30] (03PS1) 10Zfilipin: VisualEditor renamed "test" folder to "tests" [selenium] - 10https://gerrit.wikimedia.org/r/146750 [10:49:06] (03PS2) 10Zfilipin: VisualEditor renamed "test" folder to "tests"[1] [selenium] - 10https://gerrit.wikimedia.org/r/146750 [11:06:21] kart_: do you have a minute for mr question? [11:08:53] kart_: nevermind, figured it out [11:41:26] hello zeljkof [11:41:32] hi vikas_ [11:43:08] zeljkof: ah late. Was on other tab :) [11:44:16] kart_: I do have bash question, if you have a question :) [11:44:55] for example [11:44:56] https://github.com/zeljkofilipin/dotfiles/blob/master/.mrconfig#L5 [11:45:24] I would like to change this line to delete all branches except cloudbees and master [11:46:07] I thought something like this would do it [11:46:13] git branch | grep (cloudbees|master) | xargs echo [11:46:21] but I get [11:46:22] -bash: syntax error near unexpected token `cloudbees' [11:47:38] looking.. [11:49:56] zeljkof: that is correct. [11:50:04] wait. [11:50:31] kart_: my bash-fu is not there yet :( [11:52:45] (03PS1) 10Zfilipin: VisualEditor renamed "test" folder to "tests"[1] [integration/jenkins-job-builder-config] (cloudbees) - 10https://gerrit.wikimedia.org/r/146758 [11:54:28] * zeljkof is out to lunch [12:19:08] kart_: I am back, in case you have commented, I did not see it, my client disconnected [12:52:45] zeljkof: if you are free, can you review this => https://gerrit.wikimedia.org/r/#/c/142990/ [12:53:11] vikas: sure [12:53:12] in a minute [12:53:18] ok thank you :) [13:11:31] zeljkof: what will be upload rate of the bot in commons ? [13:12:04] I do not know [13:12:08] why do you ask? [13:12:24] I have to put it in request for getting bot permission [13:13:05] well, you can make it what ever you want, right? [13:13:13] what would make sense? [13:13:30] how exact the upload rate has to be? [13:13:36] I had put 1500 uploads per minute [13:13:44] https://commons.wikimedia.org/wiki/Commons:Bots/Requests/LanguageScreenshotBot [13:13:49] wow [13:13:53] you think so? [13:13:55] so much? [13:14:03] we are running all the languages in parallel [13:14:45] It will be (30 * 50) [13:14:54] 30 screenshots for 50 languages [13:19:21] zeljkof: how many languages can be run in parallel ? can wikimedia integration handle 50 ? [13:20:16] I do not know [13:20:18] probably [13:20:27] that is a question for hashar [13:20:51] are we going to run all 50 languages immediately? [13:21:32] wait a second, I will get you the exact number [13:23:55] I might be 35-40 I think, as these many languages have translation above 85 % [13:24:19] ok [13:24:38] if we take 90 and above it comes to 35 [13:24:42] I am not sure what to tell you, never applied for a bot on production [13:24:42] *90% [13:24:50] I am getting a lot of failing tests when running 142990 [13:25:36] 7 failed scenarios, 6 of them with "timed out after 5 seconds" [13:25:52] :( and another 1 ? [13:27:25] I am running all of them again in my machine, let me see [13:33:58] vikas: the next test run: 21 scenarios (5 failed, 16 passed) [13:35:00] zeljkof: Is it happening for english language ? or other language ? [13:35:07] for hr [13:35:17] switched to en, to make sure that is not the problem [13:35:19] trying now [13:39:09] zeljkof: I am getting same result as yours in english, let me check [13:41:11] zeljkof: wow ! lot of changes happened in VisualEditor :( [13:49:48] what changed? [13:50:37] Apply button/Save Page button is the dialog box is shifted to top [13:52:36] vikas: ok, so that is the problem? [13:52:38] can you fix it? [13:52:59] zeljkof: yup I am doing that [13:53:09] vikas: great, thanks, ping me when done :) [14:09:43] looks like they changed some class names also :( [14:36:04] zeljkof: OK to merge https://gerrit.wikimedia.org/r/#/c/146174/ and https://gerrit.wikimedia.org/r/#/c/146173/ ? that will make a difference in some failure in other builds [14:36:24] chrismcmahon: working on that right now [15:02:38] chrismcmahon: how do I know if a page is protected? [15:03:45] reading this, but it does not say https://www.mediawiki.org/wiki/Help:Protected_pages [15:03:50] https://www.mediawiki.org/wiki/Help:Protecting_and_unprotecting_pages [15:05:45] zeljkof: actually, I don't know how to tell without actually trying to edit the page [15:06:07] chrismcmahon: so the only way is to create another user and try to edit the page? [15:07:08] zeljkof: I don't know. there might be another way, I haven't looked. [15:12:23] am I doing something wrong? [15:12:32] client.protect_page("Title", "reason") [15:12:34] and I get [15:12:40] "{\"error\":{\"code\":\"noprotections\",\"info\":\"The protections parameter must be set\"}}", [15:13:57] reading https://www.mediawiki.org/wiki/API:Protect [15:14:40] aha [15:14:41] protections: A pipe-separated list of protections, formatted as action=group, e.g. edit=autoconfirmed|move=sysop [15:15:33] chrismcmahon: what does MF need? [15:15:43] is this ok? edit=autoconfirmed|move=sysop [15:18:16] zeljkof: hmm, OK. I did it that way because when I am logged in as Selenium_user and I go to a page, click More/Protect, the link address I get is "http://en.wikipedia.beta.wmflabs.org/w/index.php?title=0.9753751036924678&action=protect" [15:19:12] zeljkof: let's make your update "edit=sysop|move=sysop" just to be safe [15:28:09] chrismcmahon: ok, will try [15:29:55] chrismcmahon: ok, so when the page is protected, "Edit" link becomes "View source" [15:30:58] zeljkof: Assuming you can't editing through the protection. [15:31:24] JohnLewis: yes [15:31:28] zeljkof: that should be OK. the only target page for this is "Selenium Protected Page 2" or some such [15:31:49] (03CR) 10Jforrester: [C: 032] VisualEditor renamed "test" folder to "tests"[1] [selenium] - 10https://gerrit.wikimedia.org/r/146750 (owner: 10Zfilipin) [15:31:51] (03Merged) 10jenkins-bot: VisualEditor renamed "test" folder to "tests"[1] [selenium] - 10https://gerrit.wikimedia.org/r/146750 (owner: 10Zfilipin) [15:31:55] chrismcmahon: and history says: (Protected "Title": reason ([Edit=Allow only administrators] (indefinite) [Move=Allow only administrators] (indefinite))) [15:32:12] zeljkof: works for me [15:34:47] (03PS7) 10Zfilipin: Added Client#protect_page method [ruby/api] - 10https://gerrit.wikimedia.org/r/146174 (owner: 10Cmcmahon) [15:34:53] zeljkof: aha, I just noticed that there is another step past clicking Protect in the UI, which is to select a level. I think what you have is fine. [15:35:48] (03CR) 10Zfilipin: [C: 032] Added Client#protect_page method [ruby/api] - 10https://gerrit.wikimedia.org/r/146174 (owner: 10Cmcmahon) [15:35:50] (03Merged) 10jenkins-bot: Added Client#protect_page method [ruby/api] - 10https://gerrit.wikimedia.org/r/146174 (owner: 10Cmcmahon) [15:45:42] gah. why is beta db in readonly mode? [15:45:44] (03PS1) 10Zfilipin: Minor style cleanup. Updated documentation with new features. [ruby/api] - 10https://gerrit.wikimedia.org/r/146794 [15:47:40] chrismcmahon: I have to go now, mediawiki-api commit is merged, will test later mediawiki-selenium and MF commits and merged them into master [15:47:50] thanks zeljkof ! [15:48:25] chrismcmahon: I was hoping to complete all merges today, but I ran out of time :( [15:53:25] (03CR) 10Manybubbles: [C: 032] "Makes sense to me." [ruby/api] - 10https://gerrit.wikimedia.org/r/146794 (owner: 10Zfilipin) [17:12:21] 3Wikimedia / 3Continuous integration: Alert when time to merge (from +2 in Gerrit to merged in git) exceeds a known bad limit - 10https://bugzilla.wikimedia.org/68113 (10Greg Grossmeier) 3NEW p:3Unprio s:3normal a:3None Exact limit we can bikeshed/adjust as needed, but, alerting on the fact that merg... [17:15:05] 3Wikimedia / 3Continuous integration: Alert when time to merge (from +2 in Gerrit to merged in git) exceeds a known bad limit - 10https://bugzilla.wikimedia.org/68113 (10Greg Grossmeier) [17:15:07] 3Wikimedia / 3Continuous integration: Track and graph mean time to merge - 10https://bugzilla.wikimedia.org/68114 (10Greg Grossmeier) 3NEW p:3Unprio s:3normal a:3None We should track the mean time to merge, probably per repo?, so we can see trends and the effects of changes to eg test coverage or har... [17:15:20] 3Wikimedia / 3Continuous integration: Alert when time to merge (from +2 in Gerrit to merged in git) exceeds a known bad limit - 10https://bugzilla.wikimedia.org/68113 (10Greg Grossmeier) p:5Unprio>3Normal [17:15:20] 3Wikimedia / 3Continuous integration: Track and graph mean time to merge - 10https://bugzilla.wikimedia.org/68114 (10Greg Grossmeier) p:5Unprio>3Normal [17:16:07] 3Wikimedia / 3Quality Assurance: experiment with adding back HTTP::Persistent - 10https://bugzilla.wikimedia.org/68115 (10Chris McMahon) 3NEW p:3Unprio s:3normal a:3None Now that we don't have to deal with Cloudbees, investigate using HTTP::Persistent again to help performance. [17:19:33] (03CR) 10Dduvall: [C: 04-1] "Left some comments related to the in-progress Ruby coding conventions[1]—which I would love your feedback on, BTW!" (032 comments) [ruby/api] - 10https://gerrit.wikimedia.org/r/146794 (owner: 10Zfilipin) [18:07:00] why is the test failing? why iss the test failing? oh, that's a real bug! https://bugzilla.wikimedia.org/show_bug.cgi?id=68116 [18:22:00] Hello James_F [18:22:11] James_F: There are certain strings like in => https://www.mediawiki.org/wiki/File:VisualEditor_-_Media_editing_6.png which needs to be translated for different languages screenshots in user guide [18:24:08] James_F: so should I add those strings to "modules/ve-mw/i18n/en.json" in order to get it integrated with translation? [18:35:28] vikas: You mean the demonstration content? [18:35:40] James_F: yup [18:35:59] vikas: I'm not sure what the best thing to do is; getting translators to write custom content (including creating pages on the wiki) seems a bit odd. [18:37:59] James_F: hmm ! ya until we make the custom content somehow related to tool ? (but it might make User Guide confusing) [18:38:29] vikas: aharoni might have a better plan for this. :-) [18:39:09] i already spoke a bit about this with vikas [18:39:38] I think that adding an i18n with a few more messages file and making it translatable in translatewiki makes sense [18:40:04] the test scenarios and step definitions are in the same repo, and they can easily read the json files [18:40:07] aharoni: Really? Writing multiple fragments of articles sounds like a real mess. [18:40:26] it's not that much [18:40:32] very few strings, if I understand correctly [18:40:35] Right now it's "just" one dialog. [18:40:37] and some even should remain in English [18:41:06] like image search... cause commons is not *really* internationalized [18:41:24] https://www.mediawiki.org/wiki/File:VisualEditor_-_Media_editing_6.png alone is one paragraph with and annotations, a link target and the concept of what pages do and don't exist on the wiki. [18:41:29] This is going to be insane. [18:41:45] A huge amount of work. [18:42:08] Can we do something simpler? [18:43:46] it's a short paragraph [18:43:51] there are much worse messages to translate [18:44:28] It's the start of a huge amount of work. [18:44:54] and pages can be created using the testing API [18:45:03] No, they can't. [18:45:11] This has to run in a production context to work. [18:45:28] Much of the interface is wiki-specific. [18:45:30] drat [18:45:34] Yeah. [18:45:49] I'll have to think about it. [18:45:50] Hence why faking the API responses for the screenshot could be a real mess. [18:45:52] Sorry. [20:55:15] chrismcmahon or anyone, three of our browsertests on beta labs 2 hours ago failed with "getaddrinfo: Name or service not known (SocketError) [21:01:33] 90 second timeout with no new command sent to SauceLabs https://integration.wikimedia.org/ci/view/BrowserTests/job/browsertests-VisualEditor-en.wikipedia.beta.wmflabs.org-linux-firefox-sauce/lastCompletedBuild/testReport/(root)/VisualEditor%20general%20text%20markup%20features/VisualEditor_general_markup__outline_example_____Superscript____sup_General_Markup_VisualEditor_Test__sup____/ [21:30:37] 3Wikimedia / 3Quality Assurance: browser tests failing with "getaddrinfo: Name or service not known (SocketError)" - 10https://bugzilla.wikimedia.org/68125 (10spage) 3NEW p:3Unprio s:3major a:3None This was mentioned on the "[QA] Failing MobileFrontend browser tests" thread in July. I have seen a few... [21:31:00] chrismcmahon: ^ I filed a bug, will send e-mail to the QA thread that mentioned this. [21:31:27] spagewmf: thanks, I'll see if I can track it down. probably a Jenkins performance issue [21:33:49] 3Wikimedia / 3Quality Assurance: browser tests failing with "getaddrinfo: Name or service not known (SocketError)" - 10https://bugzilla.wikimedia.org/68125#c1 (10Chris McMahon) In at least one case this was caused by SauceLabs not receiving any command for 90 seconds: https://integration.wikimedia.org/ci/vie... [21:35:35] 3Wikimedia / 3Quality Assurance: browser tests failing with "getaddrinfo: Name or service not known (SocketError)" - 10https://bugzilla.wikimedia.org/68125#c2 (10Chris McMahon) Seems to be a 90 second timeout for a Flow job as well, but without the big red error message at the top of the Sauce page: https://... [22:06:13] chrismcmahon: looking at your bug 58125 comment #2 Watir GETs element/6/name "button", then there's a POST frame id: null, then 11 seconds later it ends with the screenshot [22:07:52] but if you compare that Firefox Saucelabs failure with a successful one like https://saucelabs.com/jobs/d8cdea3da354475f8007b3b8ff7c6f83 at around 40 seconds, the latter does the POST frame id: null, then immediately POST element and GETs the button again, and continues. [22:09:40] spagewmf: see https://saucelabs.com/jobs/9e6a55d3a7554a7f9a67d53dbf0f328b on the next-to-last step the "GET screenshot" is 51s (+0.90s) [22:10:04] spagewmf: Sauce took the screenshot before it killed the VM at +90s [22:10:35] but 0.90s != +90s :) [22:12:28] spagewmf: but 40s + 51s > 90s it looks like [22:13:15] I could be wrong but that's how I read it [22:13:16] seems to me like Sauce Labs hangs around waiting for instructions from Watir, and after 11 seconds it takes a screenshot and shuts down [22:14:33] the screencast movie is only 66 seconds [22:17:24] I read that as 1) post Frame, nothing happens, then 2) take a nice screen shot for the customer and kill the VM. If I were Sauce I wouldn't archive video where nothing happens. Again, I could be wrong. [22:19:06] spagewmf: meta is that I think the Jenkins host needs more juice to support the performance we need, or else we need to throttle the number of simultaneous builds. [22:19:47] and I would prefer to shore up the Jenkins host [22:20:56] chrismcmahon: that couldn't hurt :) It would be interesting to correlate watir timeouts with the getaddrinfo failures and compare with Jenkins load to see if they're both due to load. [22:22:11] chrismcmahon: I've kicked off new browser test builds on integration.wikimedia.org after a transient failure, should I not do that to avoid overloading Jenkins ? [22:23:10] spagewmf: actually, I bet the build you just started will be much improved. it's when we kick off all the builds at once starting at iirc 11AM that we suffer. [22:23:59] OK thx. I'd hate to screw up someone else's jobs while hoping mine go green :) [22:28:44] spagewmf: so just FYI on Cloudbees we had a hard limit of 2 Jenkins build executors. On WMF Jenkins we have as many as we can launch up to 10, so we're creating a LOT of load that wasn't there before. [22:42:30] spagewmf: do you know about the color console output? https://integration.wikimedia.org/ci/view/BrowserTests/job/browsertests-Flow-en.wikipedia.beta.wmflabs.org-linux-chrome-sauce/66/console [22:42:55] it's pretty nice when you're monitoring a build closely [22:44:20] we have a hung job in Zuul: https://integration.wikimedia.org/zuul/ - the one in the "Check" queue [22:44:54] I've commented on the change with "recheck" and that worked (it spawned a new job, and it completed correctly), but that one still just sits there and stares at me, laughing [22:58:53] spagewmf: looks like real failures: https://integration.wikimedia.org/ci/view/BrowserTests/job/browsertests-Flow-en.wikipedia.beta.wmflabs.org-linux-chrome-sauce/lastCompletedBuild/testReport/(root)/Topic%20collapsing/Small_topics_view/ [22:59:18] thanks, I'll take a look [23:01:15] chrismcmahon: any idea on kicking Zuul? [23:02:08] greg-g: erm, never done it. Reedy restarted Jenkins earlier today, I've never done that either, not sure I even have access to gallium (or whatever the host is) [23:05:08] yeah, apparently you can restart Jenkins via the web interface now, not sure about Zuul /me looks... nope, and it warns to not take restarting lightly because any jobs gerrit sends during that time will be lost