[00:05:24] !log tools Depooling tools-sgewebgrid-lighttpd-0924 (T244791) [00:05:27] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [00:05:27] T244791: Scale up 2020 Kubernetes cluster for final migration of legacy cluster workloads - https://phabricator.wikimedia.org/T244791 [00:05:43] !log tools Depooling tools-sgewebgrid-lighttpd-0923 (T244791) [00:05:46] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [00:08:07] bd808: Did you ever get any delete options to work for tool records yet? [00:08:32] DSquirrelGM: no, that is not currently anything I am working on [00:09:14] !log tools Depooling tools-sgewebgrid-lighttpd-0922 (T244791) [00:09:17] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [00:14:05] !log tools Depooling tools-sgewebgrid-lighttpd-0921 (T244791) [00:14:08] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [00:14:08] T244791: Scale up 2020 Kubernetes cluster for final migration of legacy cluster workloads - https://phabricator.wikimedia.org/T244791 [00:19:41] !log tools Depooling tools-sgewebgrid-generic-0904 (T244791) [00:19:44] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [00:19:44] T244791: Scale up 2020 Kubernetes cluster for final migration of legacy cluster workloads - https://phabricator.wikimedia.org/T244791 [00:20:47] !log tools Depooling tools-sgewebgrid-generic-0903 (T244791) [00:20:50] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [00:44:52] !log tools.templatehoard Manually fixed service.manifest which was broken by recent webservice changes [00:45:05] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.templatehoard/SAL [00:46:05] !log tools.antigng-bot Manually fixed service.manifest which was broken by recent webservice changes [00:46:06] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.antigng-bot/SAL [00:55:26] I'm trying to learn more about toolforge. what "are" tools exactly? are they just scripts in some language hosted on wikimedia servers? [00:55:29] what are they allowed to do? [00:56:04] ningu, that's a good question and it might not be an easy answer [00:56:15] ningu, see https://wikitech.wikimedia.org/wiki/Portal:Toolforge ? [00:56:37] andre__: thanks, I found other pages but not that yet [00:57:06] hmmm. https://wikitech.wikimedia.org/wiki/Help:Glossary#Toolforge - Tool - Proper name for a bot/webservice/job that runs in Toolforge. [00:57:10] https://wikitech.wikimedia.org/wiki/Portal:Toolforge/About_Toolforge#What_is_a_tool? and https://wikitech.wikimedia.org/wiki/Portal:Toolforge/About_Toolforge#What_is_Toolforge? [00:57:25] https://meta.wikimedia.org/wiki/Tools [00:57:37] my specific question is, basically, is it possible for a logged in wikipedia user to click a link to a toolforge tool and have that tool make edits under their credentials (with their permission if necessary, of course)? [00:57:53] yes [00:58:11] also can the tool receive query params or similar info within the link? [00:58:12] you can run a webservice in tools and do some oauth thing with wikipedia [00:58:24] ok cool, that sounds typical [00:58:31] I mean typical way to do these things [00:58:46] yes [00:58:50] and yes [00:58:52] thanks [00:59:02] ningu: https://wikitech.wikimedia.org/wiki/Help:Toolforge/My_first_Flask_OAuth_tool might give you some ideas about using OAuth [01:05:42] the tool I am imagining would basically be a react app front-end for editing a particular sort of page on wikisource. actually it can be compiled to javascript, maybe with just a simple html template where the react app would be mounted [01:05:57] the react app already exists as part of a mediawiki extension but it's been suggested it would make more sense as a tool [01:06:15] can that sort of thing be hosted? it wouldn't even need a running process, as long as I can update the compiled code somehow [01:06:31] maybe need process for oauth, dunno [01:07:23] ningu: yes, that would be a good fit for a webservice running on Toolforge. You would probably need to proxy the edits through a small process to manage the OAuth. Our OAuth support is not ideal for client side deployments. [01:07:48] ok no problem [01:08:06] ningu: see https://wikitech.wikimedia.org/wiki/Help:Toolforge/Web for information on serving web files in various ways [01:08:09] should the process also be a little web server serving the js? [01:08:10] ok thanks [01:08:38] nodejs, python3, and php are all pretty well supported [01:09:09] other language runtimes are available as well with various levels of documentation and ease of use [01:10:11] likely to use nodejs although python is also possible [01:10:29] another thing we need to figure out is how to implement automatic transliteration from balinese script to latin. currently using this: https://github.com/longnow/icu-transliterator-service [01:10:46] which is a node module that compiles against ICU. but apparently wikimedia has its own system for this? [01:11:56] I don't know if we have a service, you might be thinking of LanguageConverter? [01:12:04] yeah that, thanks [01:12:29] that's part of MediaWiki Core [01:13:41] IIRC it's buried in the wikitext parser? [01:13:52] it may or may not be helpful to your use case [01:14:47] ok [01:14:51] I can't quite find docs on it [01:14:55] I found this https://www.mediawiki.org/wiki/Writing_systems#LanguageConverter [01:15:01] but it doesn't really show the relevant info [01:15:29] this might be one area of Wikimedia that is better documented in languages other than English [01:16:03] ok I kind of see [01:16:07] Kazakh is here https://phabricator.wikimedia.org/source/mediawiki/browse/master/languages/converters/KkConverter.php [01:16:22] https://www.mediawiki.org/wiki/Writing_systems/Syntax [01:19:16] yeah, I want to see how the transliteration rules are actually written [01:19:36] oh, well, yeah you'll want the MW source code you already found [01:20:26] I kind of see... I dunno if this system will work, we need context-sensitive rules [01:21:15] looks like regex match is possible, in which case it will be fine [01:22:00] Technically, you can do whatever you want [01:22:03] I need to get some sleep [01:22:04] yeah I see [01:22:08] good luck [01:22:10] ok, thanks for the help! [01:22:22] ningu: are you working on T244796 ? [01:22:23] T244796: moving palmleaf.org platform to Balinese Wikisource/Wikipedia - https://phabricator.wikimedia.org/T244796 [01:22:29] bd808: yes [01:22:44] cool. Lots of interesting things to explore :) [01:22:48] definitely, yeah [01:23:53] is there further documentation on the nodejs platform for toolforge? like, can we install icu and link the required node module against it? [01:23:59] the idea of trying to make things work as a tool rather than an extension is a good one I think. Getting a new extension that needs supporting services deployed into Wikimedia's production wiki cluster is possible, but not quick or easy [01:24:06] maybe I should just make a toolforge account and fool around [01:24:20] we don't _have_ to use ICU but it's what we already have working [01:24:58] yeah, getting into the platform and experimenting is a good plan. Let me check to se if icu dev libraries are installed in our current node10 Docker image... [01:25:28] any old icu is probably fine. we have our own transliteration ruleset, so it won't be preloaded in any case [01:25:37] not sure on icu version requirement [01:28:50] ok, well I submitted a toolforge membership request [01:32:11] ningu: https://toolsadmin.wikimedia.org/tools/membership/status/709 is approved. You will need to log out of toolsadmin.wikimedia.org and log back in before you can create a tool. (its a bug) [01:35:27] thanks! [01:36:37] hmm, it says "connect your wikimedia account with your developer account"... does that mean my regular account for editing wikipedia? [01:37:13] ningu: yes [01:38:50] ok, all seems to be working [01:40:01] "create or join a tool"... I have to admit, there's something really weird in english about the idea of joining a tool, but whatever :P [01:40:24] When Metaphors Collide [01:50:21] It's possibly better than "becoming a tool" [01:53:33] i'm sure that one's snuck into documentaion or bug reports at least once [01:54:20] You literally type `become my-tool-name` so yeah :) [01:57:01] haha [02:49:26] what exactly does this command do: webservice --backend=kubernetes node10 shell [02:49:37] it returns me to the command prompt [02:50:15] saying, Pod is not ready in time [02:59:02] try it again [02:59:46] it starts a docker container I guess? [03:00:11] same result [03:00:52] yeah, it should start a docker container and then open a shell in it [03:01:16] Issues on the first run aren't unexpected, but it should work the second time [03:01:34] ok, I'll try it again on the same login session [03:02:53] same results [03:04:17] ugh who's loading down tools-sgebastion-07 again [03:04:20] bd808: shouldn't a client-side tool be doable now that we support OAuth2? [03:04:46] (re earlier discussion about a react tool) [03:08:56] yeah, I'm not able to start a node10, nodejs, or python3.7 shell [03:09:26] AntiComposite: ok well glad it's not just me, it's no problem for now [03:10:23] https://www.oauth.com/oauth2-servers/access-tokens/authorization-code-request/ is specifically meant for web clients AIUI [03:38:30] just connected and restarted a webservice on login tools without an issue now - try again? [03:38:58] seems to be running normally here [03:42:45] webservice or shell [03:43:00] because shells are still timing out for me [03:49:25] was connecting via ssh and restarted the webservice [03:49:51] seemed to be running normally there [03:50:18] on toolforge though, not vps [03:50:32] though it's the same wrapper, the way the webservice shell and webservice {start|stop|restart} commands work is not the same [04:21:28] AntiComposite: is a shell actually needed? or can you just cd to the www/js dir and run npm install? [04:21:34] and then start the service [04:21:41] I don't know [04:22:04] I would think the shell is more involved than just a cd, but who knows [04:22:15] npm exists without the shell [04:22:29] I know it's important to install Python virtual environmanets from within k8s so that you get the right versions of everything [04:22:46] but I try to avoid js as much as possible, so I don't have any experience with node :) [04:23:00] haha ok [04:23:16] and yeah, good point, the node/npm may well not be the same outside the shell [04:23:28] i get it now [04:23:57] yeah, you'll probably get node 6 or node 8 on the bastion, while k8s is node10 [04:24:07] k8s=kubernetes [04:49:14] ningu, https://phabricator.wikimedia.org/T244954 [05:09:29] thanks! [13:38:30] !log admin [codfw1dev] add reference to subnetpool to the instance subnet `MariaDB [neutron]> update subnets set subnetpool_id='d129650d-d4be-4fe1-b13e-6edb5565cb4a' where id = '7adfcebe-b3d0-4315-92fe-e8365cc80668';` (T244851) [13:38:33] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Admin/SAL [13:38:33] T244851: Neutron: replace NAT customization with address scopes - https://phabricator.wikimedia.org/T244851 [14:32:05] !log tools.wikibase-databridge-storybook cleaned up tool (rm -rf a bunch of stuff) for T240340 and T245026 [14:32:10] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.wikibase-databridge-storybook/SAL [14:32:10] T240340: Host bridge storybook on doc.wikimedia.org - https://phabricator.wikimedia.org/T240340 [14:32:10] T245026: Archive/delete tool wikibase-databridge-storybook - https://phabricator.wikimedia.org/T245026 [15:28:48] !log ores staging ores-wmflabs-deploy:4f0a14e [15:28:50] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Ores/SAL [16:45:09] Cyberpower678: Have you seen my latest comment on T241493? I think I found a pile of data you can purge to get ~145G of space back for your db. [16:45:10] T241493: cyberbot-db-01.cyberbot.eqiad.wmflabs is running out of space and needs more - https://phabricator.wikimedia.org/T241493 [19:15:10] !log tools Deployed toollabs-webservice (0.63) on bastions (T244954) [19:15:19] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [19:15:19] T244954: Webservice shell repeatedly times out - https://phabricator.wikimedia.org/T244954 [19:29:21] !log tools Rebuilding all Docker images to pick up toollabs-webservice (0.63) (T244954) [19:29:24] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [19:29:25] T244954: Webservice shell repeatedly times out - https://phabricator.wikimedia.org/T244954 [19:42:53] I am working through https://wikitech.wikimedia.org/wiki/Help:Toolforge/My_first_NodeJS_OAuth_tool [19:43:09] under step 2, it says mkdir -p $HOME/www/js but never cd's to it -- is that an error? [19:43:34] if you npm init the way they say, it will put package.json in $HOME [19:44:33] I think I'm going to assume that's an oversight :) [19:50:46] heh [19:50:50] Docs are often wrong [19:51:02] Reedy: indeed, but this is a new platform for me -- the toolforge stuff that is, not node [19:51:17] I don't see any docs on where to put the app or what the entrypoint is [19:51:23] package.json says entrypoint is index.js but that doesn't exist [19:51:59] so apparently it doesn't look to package.json for that [19:52:17] anyway I'll fix that bit in the docs [19:52:23] Expected file layout suggests package.json should be in $HOME/www/js [19:52:25] please do :) [19:52:49] do you happen to know if running webservice ... shell will stay in the same dir as where you invoked it? [19:53:00] let me check [19:53:20] it does, cool [20:00:18] Reedy: curious here. the docs say to do app.get( "/", ... ) but the actual incoming web request seems to be at //, in my case https://tools.wmflabs.org/archiveleaf/ so it comes in at /archiveleaf/ [20:00:31] wondering if that behavior has been changed? if it's supposed to be the way it is now? [20:00:53] it seems like it would be nice from a coding perspective to come in at root and not worry about it, but I dunno if it's done the way it is now for some other good reason [20:00:58] I've done minimal development with node, so I'm not really the person to be asking for it :( [20:01:03] well this is more like nginx config [20:01:04] I'd obviously suggest to test it and see if it works [20:01:19] no, this is how tools.wmflabs.org forwards the request [20:01:23] it's nothing to do with node per se [20:01:42] it may be that it forwards differently to node apps than python apps for all I know [20:01:48] but either way it's at the nginx side that does it [20:06:09] It should be setup for simplicity, so the tool behaving like it's recieving requests relative to itself (rather than relative to tools.wmflabs.org) would make sense [20:08:46] https://wikitech.wikimedia.org/wiki/Help:Toolforge/My_first_NodeJS_OAuth_tool#Update_the_application_code [20:09:01] One of the examples further on includes a line that adjusts the routing logic [20:09:25] to make Express treat // as / [20:27:24] AntiComposite: thanks [20:30:31] AntiComposite: I don't see how it really does that though [20:31:52] oh I see now [20:36:17] now I'm stuck at the register oauth consumer step. is that my wikitech account or wikimedia? [20:36:37] it seems to only accept my wikimedia account credentials, but then it says I need to verify my email address to proceed, even though I've already done that [20:39:29] ok I see, "autoconfirmed status on meta-wiki is automatically obtained after 4 days and 5 edits" [21:12:12] node10 shell stopped working again [21:12:43] ningu: what error are you getting? [21:12:54] "Pod is not ready in time", same as yesterday [21:13:07] running webservice --backend=kubernetes node10 shell [21:14:24] It just worked for me, but... there is a running process updating the Docker images. We have an "always pull" policy set, so there may be some timeouts as the various Kuberenetes work nodes pull new images from the repo. [21:14:46] I am pretty sure that the logic bug in the webservice command itself is fixed now [21:14:49] it was working a little while ago [21:15:15] ok now I got, [21:15:15] pykube.exceptions.HTTPError: object is being deleted: pods "interactive" already exists [21:15:39] ok. that sounds like the image pull caused delay [21:15:44] ok [21:15:59] take a break, grab a beverage, try again in a few mintues :) [21:16:04] sure [21:47:00] !log tools.zppixbot ZppixBotWiki to read only and begin PHP Upgrade — T240898 [21:47:03] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.zppixbot/SAL [21:47:03] T240898: Update MediaWiki + components on ZppixBot Wiki - https://phabricator.wikimedia.org/T240898 [21:53:23] bd808 or anyone else who knows -- are there logs from the webservice while running? [21:54:06] also, the shell still isn't working [21:59:39] !help php -v reports 7.2.26 but https://tools.wmflabs.org/zppixbot/wiki/index.php?title=Planning/2020 shows otherwise [21:59:39] RhinosF1: If you don't get a response in 15-30 minutes, please create a phabricator task -- https://phabricator.wikimedia.org/maniphest/task/edit/form/1/?projects=wmcs-team [22:01:24] RhinosF1: I would do webservice stop and then webservice --backend=kubernetes php7.3 start [22:01:57] Per https://wikitech.wikimedia.org/wiki/Help:Toolforge/Kubernetes#PHP [22:02:14] !log tools.zppixbot restart webservice as php7.3 [22:02:16] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.zppixbot/SAL [22:02:31] does no one use gridengine? [22:03:25] I don't even know what it is, so... :P [22:03:29] MacFan4000: now it's 502ing [22:03:44] Hmmm It works for me [22:03:51] MacFan4000: one sec [22:04:00] MacFan4000: back [22:04:04] I see, it's an oracle thing [22:07:17] MacFan4000: now, is there an easy way to update mariadb? [22:07:40] I’m not really sure how possible that is on toolforge [22:08:26] I'm going to try upgrading mediawiki without then [22:08:40] ningu: the gird engine is used quite heavily for continuous bots and cron jobs, but it is generally a less featureful solution for webservices [22:08:59] bd808: cool thanks [22:09:08] some details at https://wikitech.wikimedia.org/wiki/Help:Toolforge/Grid [22:09:13] bd808: is there any way of passing config to a web service via env vars? or only via a config file for example? [22:09:21] other than the PORT [22:09:50] there is no built in .env type solution [22:09:57] ok [22:10:27] !log tools.zppixbot Upload and prep mediawiki for T240898 [22:10:29] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.zppixbot/SAL [22:10:30] T240898: Update MediaWiki + components on ZppixBot Wiki - https://phabricator.wikimedia.org/T240898 [22:11:11] using webservice to run things is a lot of convention over configuration. Probably too much, but changes have to be made a bit slowly so that we don't break the ~1200 webservices that are in active use [22:11:18] FYI I believe Zpp_ixBot wiki uses git [22:12:08] bd808: ok. is there any way to detect the current tool name, then? [22:12:19] so I can mount the expressjs router at the right url? [22:12:22] MacFan4000: docs are not the wiki [22:12:54] bd808: also, should I listen on localhost or 0.0.0.0? [22:12:56] RhinosF1: I mean it was done by git cloning the MediaWiki repo [22:13:37] ningu: the current user should be the tool's user which would provide the tool name (tools. would be the username) [22:13:48] RhinosF1: as in https://github.com/wikimedia/mediawiki [22:14:03] ningu: for a nodejs service I would build to 0.0.0.0 [22:14:08] *bind to [22:14:14] thanks [22:15:09] the actual Pod will only get traffic brokered by its associated Ingress object. These words are gibberish though if you are not familiar with Kubernetes internals [22:15:35] MacFan4000: you might be right [22:16:15] bd808: yeah no worries, I just know from a bit of docker experience that to get outside a container you generally need to not use localhost but some outside interface [22:16:22] traffic from the outside world goes something like: [internet] -> [tools.wmflabs.org nginx proxy] -> [Kubernetes ingress] -> [namespace ingress] -> [namespace service] -> [pod] [22:16:45] but indeed I don't know much about kubernetes other than in general what it is [22:17:41] MacFan4000: it's doing something [22:17:52] just slow [22:22:57] bd808: any idea why node10 shell is still not working? [22:24:07] I just started one myself after you asked, so no. If you would like to share your tool name I can use maigc superpowers to try it as that tool and see if I can figure something out. [22:24:30] icu-transliterate [22:25:09] I am working from the 2020 Kubernetes cluster rather than the legacy cluster. That might make some difference actually. I will test from your tool [22:25:27] I'm connecting to tools-login.wmflabs.org, is that ok? [22:25:47] there is also tools-dev or something that I saw? wasn't sure about that [22:25:48] yes, that's the right place. The 202 cluster stuff is about https://wikitech.wikimedia.org/wiki/News/2020_Kubernetes_cluster_migration [22:25:57] *2020 cluster [22:26:00] ok [22:26:13] I see, it's already underway? [22:26:38] yeah, we are about a week from force migrating everyone [22:26:46] I can manually migrate [22:27:10] is there any difference between login.tools. and tools-login. ? [22:27:17] well it says no webservice running [22:27:21] so maybe I just need to alias kubectl? [22:27:40] no webservice is running cause I haven't been able to open a shell, which I need to npm install ... [22:28:02] DSquirrelGM: nope. there are at least 3 names in DNS for the same hosts. The soon to be canonical name is actually login.toolforge.org [22:29:20] is that completely set up yet? if so I'll start using that for my ssh - less typing [22:29:32] I aliased mine to just "tools" anyway :P [22:29:44] DSquirrelGM: yes, the login.toolforge.org DNS name is active. [22:30:41] ningu: I can confirm the node10 shell startup problem from your tool. I will keep digging a bit, but may just switch your config to the 2020 cluster [22:31:08] bd808: can I do that myself? [22:31:54] ningu: yes. you need to do steps 5 & 6 from https://wikitech.wikimedia.org/wiki/News/2020_Kubernetes_cluster_migration#Manually_migrate_Kubernetes_continuous_jobs_or_other_custom_Deployments [22:32:05] ok [22:32:31] !log tools.zppixbot upgrading using git pull && update.php to each major version in turn + composer [22:32:32] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.zppixbot/SAL [22:32:34] the alias weirdly seems not to be working [22:32:38] ok just connected and added to my known hosts [22:32:48] there's a difference between /usr/bin/kubectl version and kubectl version [22:34:02] seems ok now, with .bashrc instead of .profile [22:34:29] RhinosF1: you’ll have to be sure that the submodules are up to date as well (skins/extensions/vendor) [22:34:49] MacFan4000: it was doing it automatically [22:34:59] waiting on composer now [22:35:27] anyway, seems to be working now [22:38:02] ningu: glad that unblocked you. I'm going to keep looking at the legacy cluster to see what's wrong there. [22:42:30] !log tools.zppixbot 1.31.6 - 1.32 latest commit failed - trying using git reset --hard and hoping [22:42:32] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.zppixbot/SAL [22:49:57] bd808: are you around to debug me causing a HTTP 500 on https://tools.wmflabs.org/zppixbot/wiki/index.php?title=Special:Version [22:50:15] it works fine [22:50:20] works for me? [22:50:57] https://tools.wmflabs.org/zppixbot/wiki/index.php?title=Planning/2020 [22:51:18] bd808: seems to have been slow to restart when I upgraded [22:52:10] next jump [22:53:17] are tools on toolforge meant to be usable from client-side browsers? i.e. with CORS? [22:54:46] ningu: that is up to the tool, but I think we add a CORS header to all requests at the front proxy. [22:54:58] ok, but it isn't discouraged, is all I really mean [22:55:01] and is possible [22:55:28] yes and yes. A number of tools are backends for on-wiki gadgets [22:55:33] cool [22:56:54] Reedy: I'm now going to update when you email the tool's dev wiki so I never have to do this again [22:58:58] for ningu and others, I debugged the `webservice --backend=kubernetes node10 shell` issue on the legacy kubernetes grid and found that for currently unknown reasons it is taking ~90 seconds for new pods to start. That is longer than the currently hardcoded wait time for the shell action. [22:59:37] I'm going to say that the "fix" is to use the 2020 cluster. That's a lame answer, but expedient [23:01:55] no problem [23:02:12] easy enough to do [23:02:39] hrm... do I have to switch context each time I log in? [23:02:46] maybe not [23:03:03] ningu: no, that change is persisted in a dot file [23:03:42] specifically $HOME/.kube/config [23:03:56] bd808: is it only me that sees the wiki on 1.33.2 rendering plain text now and noot wiki-like [23:04:26] RhinosF1: yeah your resourceloader seems to be broken somehow [23:04:44] bd808: this is going 1.32.6 to 1.33.2 [23:04:48] what changed? [23:04:54] code? [23:05:33] bd808: that could cause this and how can I fix it [23:07:25] RhinosF1: this is not generally a thing I can spend the time to help you with. I would recommend reading the error logs and examining the response data sent to your browser. The output of https://tools.wmflabs.org/zppixbot/wiki/load.php?lang=en&modules=mediawiki.legacy.commonPrint%2Cshared%7Cmediawiki.skinning.interface%7Cmediawiki.special.version%7Cskins.vector.styles&only=styles&skin=vector might be interesting. [23:13:05] now my web service seems to be hanging, getting error 504 request timed out [23:13:30] gateway timeout, rather [23:14:34] ok, well I kinda see what might be going on [23:14:38] I think it's on my end :) [23:18:25] awesome, it's working [23:30:09] !log tools.zppixbot unlocked wikidb T240898 (updated most things and got it working) [23:30:11] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.zppixbot/SAL [23:30:11] T240898: Update MediaWiki + components on ZppixBot Wiki - https://phabricator.wikimedia.org/T240898 [23:30:23] bd808: needed to update exts. Fixed! [23:43:32] ok, somehow I still can't create an oauth consumer... I guess I have to wait 4 days to be approved? [23:45:13] Yeah [23:45:22] if you've made the edits - https://meta.wikimedia.org/wiki/Meta:Autoconfirmed_users [23:45:28] I have [23:46:00] ningu: what's your username? [23:46:14] btw is it possible somehow to use the static file server to instantiate a react app, and then do oauth client side? [23:46:18] or do you always need the server backend? [23:46:35] tgr: Lautgesetz [23:46:40] should be possible with OAuth 2 [23:46:53] but I don't know much about OAuth 2 [23:47:19] ok. I'll look into how it works client-side in general, I guess [23:48:06] the OAuth proposal form for OAuth 2 has a "Client is confidential [23:48:17] field, which I imagine is related to that [23:51:11] hm, looks like staff accounts cannot manage user rights [23:51:21] I'm pretty sure that was a thing at some point [23:51:36] this explains some of it: https://aaronparecki.com/oauth-2-simplified/ [23:51:43] it looks like it should be possible [23:52:41] oh, well. You'll have to ask a meta admin or steward to make you autoconfirmed, then [23:52:50] ok no worries, thanks [23:53:02] (relevant task: T213760) [23:53:02] T213760: Rethink autoconfirmed requirement for OAuth - https://phabricator.wikimedia.org/T213760 [23:54:35] would be just 'confirmed', technically [23:56:06] no, apparently I've been confirmed but not autoconfirmed [23:56:10] but apparently both are needed [23:58:16] You're not confirmed [23:58:17] https://meta.wikimedia.org/w/index.php?title=Special%3AUserRights&user=Lautgesetz [23:58:22] email confirmed != confirmed [23:58:31] https://meta.wikimedia.org/wiki/Meta:Confirmed_users [23:59:03] Reedy: all I know is, earlier today on the page at https://meta.wikimedia.org/wiki/Special:OAuthConsumerRegistration/propose, it said I was unauthorized cause i needed to be in the group autoconfirmed or confirmed, and then it just said autoconfirmed, so I assumed I had been put under confirmed [23:59:09] but now i see it's back to listing both groups, so who knows [23:59:20] You can look at https://meta.wikimedia.org/wiki/Special:ListGroupRights [23:59:28] Both autoconfirmed and confirmed have: [23:59:29] Propose new OAuth consumers (mwoauthproposeconsumer) [23:59:32] Update OAuth consumers you control (mwoauthupdateownconsumer) [23:59:44] ok