[03:00:06] [telegram] TG群发，私聊，在线，增粉；三万优质中文群全网监听服务限时特价；需要请咨询 : https://tools-static.wmflabs.org/bridgebot/d368aa44/file_1498.jpg [07:50:12] [telegram] Sam, where can I find the status page? Looking at https://admin.toolforge.org/ but it's not linked. (Montage still seems to be down) (re @wmtelegram_bot: [irc] >Montage is hosted on Toolforge, which provides the servers and databases, as well as database backups. Keep an eye on the Toolforge status page as well.) [09:42:37] [telegram] Montage and Fountain are still down. Any news on the server? [09:53:14] [telegram] migration is still ongoing as far as I’m aware – you can probably check https://lists.wikimedia.orghttps://phabricator.wikimedia.org/T266587ail/cloud-announce/2020-November/thread.html or T266587 for updates [09:53:15] T266587: ToolsDB replication is broken - https://phabricator.wikimedia.org/T266587 [12:27:14] [telegram] Hello, Anyone here can translate this sentence to Hindi or Japanese please? "Wikimedia projects including Wikipedia are getting more secure. You are using an old web browser that will not be able to connect to Wikimedia projects in the future. Please update your device or contact your IT administrator." [12:27:23] [telegram] (or ping people who can?) [12:27:33] [telegram] Context: https://gerrit.wikimedia.org/r/c/operations/puppet/+/637850 [13:01:40] [telegram] https://lists.wikimedia.org/pipermail/cloud-announce/2020-November/000341.html ( https://wikitech.wikimedia.org/wiki/News/Wiki_Replicas_2020_Redesign ) — this sounds like crappy thing. This will have huge negative influence on Quarry among other things, won't it? [13:02:23] [telegram] I definitely did write a few cross db join queries there... [13:02:57] [telegram] E.g. to find how many contribs the activemost ukwiki users have on WD and stuff [13:03:39] [telegram] You could do that in other ways that don't involve using DB queries [13:15:03] [telegram] Sure, or many DB queries, but all of those require writing some backend. No more quick answers within 10 minutes of a question being asked in a chat an idea occuring. [13:15:12] [telegram] Sure, or many DB queries, but all of those require writing some backend. No more quick answers within 10 minutes of a question being asked in a chat or an idea occuring. [13:29:19] [telegram] It was known for a while that it would come to this. Is ever-increasing replication lag a better option? (re @Thecladis: Sure, or many DB queries, but all of those require writing some backend. No more quick answers within 10 minutes of a question being asked in a chat or an idea occuring.) [13:31:58] [telegram] It might be. I am not sure if these are the only alternatives. [13:32:44] [telegram] Replication lag of several minutes is better than a programme one has to write instead and spend more time for that [13:35:24] [telegram] Pinged a volunteer from the Hindi community (re @Ladsgroup: (or ping people who can?)) [13:38:15] [telegram] Ever increasing. (re @Thecladis: Replication lag of several minutes is better than a programme one has to write instead and spend more time for that) [13:39:40] [telegram] A few minutes for a few months, tens of minutes for a few months, hours for a few months... [13:41:37] [telegram] Have we hit a ceiling that for example more resources would not fix? [13:41:49] [telegram] We do have money for strategy and other shit after all :) [13:42:48] [telegram] There were big lags back at Toolserver... [13:43:48] [telegram] Yeah, was that better? (re @Thecladis: There were big lags back at Toolserver...) [13:45:18] [telegram] It wasn't. But I am pretty sure the problem was solved with basically more funding [13:46:21] [telegram] Sure, but how much money should be invested in this use-case, instead of the tens of features we are waiting for in Wikimedia Cloud services? (re @Thecladis: We do have money for strategy and other shit after all :)) [13:48:00] [telegram] Why does it have to be an instead? [13:49:40] [telegram] Because there are limited resources in the team. And cross wiki queries are a very expensive computationally and in human resources required to support any architecture that would allow it. [13:49:53] [telegram] Because there are limited resources in the team. And cross wiki queries are very expensive computationally and in human resources required to support any architecture that would allow it. [13:51:00] [telegram] Well that computation has to happen somewhere. Either still on labs, or on volunteers' machines (which are often not exactly top-notch ones) [13:53:52] [telegram] Besides when it comes to Quarry we are talking about tons of queries that often barely use any indices and so forth. We don't have a DBA team handy to just write any queries for us, volunteers write them, often with bare minimum SQL knowledge. I am pretty sure that there are other expensive computations being made as the result [13:54:33] [telegram] Sure, but instead of one humongous db there will be 8 DBs. Adding a few expensive queries, or doing the join in application code doesn't compare to managing a db this size. [13:56:31] [telegram] It does if you add it up. Again, that computation and all other stuff does not just magically disappear. It just gets outsourced to volunteers' time, skills and sometimes machines. Or sometimes not volunteers' [13:57:32] [telegram] Many tools would just die [13:57:46] [telegram] Not all tools recovered from as simple thing as the actor migration [13:57:58] [telegram] although the benefit of that is undoubtful [14:01:30] [telegram] hmm, I think it is also past time Translate gets installed on Wikitech… [14:10:12] [telegram] thanks (re @Ciell: Pinged a volunteer from the Hindi community) [15:18:23] [telegram] If your point is WMF should invest more in teams that can either develop new tools for the wikimedia movement or pickup the maintenance for tools that are not maintained and still are useful to the movement I agree. (re @Thecladis: Many tools would just die) [15:23:54] [telegram] More or less. Going as far as developing those tools might be extreme, though for some cases where it is crucial that the tools are up to industry standards it is well justified. But in some cases it is just that the support should be even better for volunteers to make those. And that is not so much a question of money allocation within the team as within the WMF as the whole [16:23:57] [telegram] many tools are open source and under a free license, so I guess all we need is someone to manage the job... [20:55:25] [telegram] Not sure if https://wikidata-terminator.toolforge.org still works, but would be a good option to increase coverage for a language [20:57:50] [telegram] then use variant https://wikidata-terminator.toolforge.org/#/top/labels