[07:59:42] 2.3 PB/s. Suuuuuuuuure. https://ganglia.wikimedia.org/latest/?r=month&cs=1%2F16%2F2014+13%3A15&ce=1%2F18%2F2014+10%3A3&m=cpu_report&s=by+name&c=Text+caches+eqiad&h=&host_regex=&max_graphs=0&tab=m&vn=&hide-hf=false&sh=1&z=small&hc=4 [08:17:56] does anyone have anything to say about http://www.wikidata.org/w/index.php?title=Wikidata:Forum&curid=6347&diff=105433339&oldid=105418841 ? [08:18:16] maybe Gloria / Nemo_bis? [08:19:30] cstrippe is the one that handles security [08:19:52] direct them to bz, where you can do things like cc people properly and add keywords [08:20:23] there have definitely been BZs about the ciphers in the past [08:20:51] and afaik right now you can only do it wrong this way or that way currently [08:22:30] for various misc services SSLCipherSuite AES128-GCM-SHA256:RC4-SHA:RC4-MD5:DES-CBC3-SHA:AES128-SHA:AES256-SHA [08:22:43] SSLHonorCipherOrder on [08:24:24] Jasper_Deng: that comment says that they want to use a cipher because it's broken. Not very useful. [08:24:56] yea, they say when they disable the broken one they cant connect [08:25:10] but that is probably the feature, isnt it [08:25:17] Yep [08:25:24] eh, or mixed up enabled/disabled [08:25:25] Probably they meant something else [08:25:38] All we know is https://wikitech.wikimedia.org/wiki/HTTPS/Future_work but that document/job is abandoned [08:25:52] Most talk has been about RC4 https://wikitech.wikimedia.org/wiki/Talk:HTTPS/Future_work [08:26:09] that's because it was Ryan's project i supposed [08:26:35] Not really, there are a number of people with various action items, mentioned on that page [08:27:43] oh, about disabling RC4, i see, yea [08:35:15] Chris Steipp, csteipp [08:35:54] Nemo_bis: so at one point many people enabled it to mitigate beast attack. http://blogs.technet.com/resized-image.ashx/__size/550x0/__key/communityserver-blogs-components-weblogfiles/00-00-00-61-47/6153.rc4.png [08:35:59] Gloria: https://community.qualys.com/blogs/securitylabs/2013/03/19/rc4-in-tls-is-broken-now-what [08:36:28] but it's still getting an A- see https://www.ssllabs.com/ssltest/analyze.html?d=wikidata.org [08:36:43] see "RC4" comments in there [08:37:55] I'm not really following this discussion. [08:38:02] I was just correcting "cstrippe". [08:38:05] eh, then s/Gloria/Nemo_bis :) [08:38:08] ok [08:38:18] thought you were pinging him about that [08:38:31] just an e-r typo I think [08:38:53] mutante: Yes I know a bit about that, but when will we have a decision? :) [08:39:24] Same on cluster links: even Microsoft encrypts them nowadays. :P [08:41:28] not sure i know what you mean there, we offer https [08:41:40] what's not encrypted that should be [08:43:08] or do you mean enforcing it and disabling http [09:05:59] I was just correcting "cstrippe". is it eight-five-eight or eight hundred fifty-eight? [09:09:25] Heh. [09:13:01] i'm just kidding. [10:06:40] ori: i was referring to my daystar public name [10:09:21] p858snake|l: i know; i was just kidding. i didn't mean to tease. [11:14:17] X!'s edit counter looks down: http://tools.wmflabs.org/xtools/pcount/ [11:14:48] It now only shows "No input file specified" (ld?) [12:36:27] lol, someone at news.net needs to learn API, 8 millions requests per day seriously :P https://docs.google.com/document/d/1kpJrfataS5KAxGXFoygQVhMlzFftjsvX9HktSAAKfrQ/edit [12:40:36] Hm. http://www.80legs.com/services.html [14:05:11] oh, gamepedia.com uses the Drafts extension [16:25:21] the 'list articles created by this user' tool is broken. Does anyone know a) how to fix it, and/or b) to whom complaints should be addressed? [16:26:01] 1) No, 2) the author [18:21:48] :/ [18:21:54] wrong window [18:29:00] is Dan Garry here? [18:29:30] <^d> Not in this channel. [18:29:44] <^d> I pinged him for you. [18:29:53] <^d> twkozlowski: There you go ^ [18:29:58] Someone called me? :) [18:30:03] so thoughtful, ^d :* [18:33:17] ^d: manybubbles: what search deployments are we doing this week? [18:34:12] <^d> Nothing today. Maybe huwiki wednesday? [18:35:45] robla: yeah. we will do enwiki as a betafeature on Monday probably [18:35:52] beyond that, bug fixes [18:36:51] yay, let's see how many en.wiki fuzzy searches it takes to make the elastic cluster gasp [18:37:08] * ^d gives Nemo_bis some beans [18:37:35] * Nemo_bis swallows happily and woofs [18:37:45] Still empty :( https://ganglia.wikimedia.org/latest/graph_all_periods.php?c=Miscellaneous%20pmtpa&h=hume.wikimedia.org&v=823574&m=Global_JobQueue_length&r=hour&z=default&jr=&js=&st=1365625056&z=large [18:37:59] <^d> lol hume. [18:38:04] <^d> we should really move that to terbium. [18:38:25] But that's one of the nicest philophers! [18:38:54] well, *the* nicest [18:39:41] I was trying to be sober [18:40:30] Nemo_bis: it'll be a race between the train (which has a fix for ~ queries) and people with heart's full of hate [18:41:24] aww [18:41:47] look at this: http://ganglia.wikimedia.org/latest/?r=hour&cs=&ce=&m=es_fetch_time&s=by+name&c=Elasticsearch+cluster+eqiad&h=&host_regex=&max_graphs=0&tab=m&vn=&hide-hf=false&sh=1&z=small&hc=4 [18:41:57] you can see the spikes in fetch time correspond with spikes in cpu time [18:42:25] that fetch time (so far as I've always seen when I check) is ~ queries [18:42:30] well, the spikes are [18:42:34] <^d> I love our rainbow graphs. [18:42:58] it is like I have super rainbow powers [18:43:33] ♫ but pal you don't find rainbows in the bottom of a glass ♫ [18:44:16] someone was complaining about links update taking a long time on the mailing list. I'm sad about that too. [19:00:33] manybubbles: are you saying that a fuzzy search can bring the network activity from about 5 to about 10 MB/s? :o [20:24:03] gwicke: When I try to run the parsoid server locally, I get the error "Cannot find module 'optimist' at Function.Module._resolveFilename (module.js:338:15)". Any idea how to fix that? [20:24:51] gwicke: Maybe I'm missing some dependancies... [20:25:43] gwicke: Nevermind, I got it working :) [21:05:44] kaldari: and did you update docs? :) [22:15:13] sob, labs still/again down? [23:19:23] MatmaRex Nemo_bis quiddity [23:19:37] Couple questions about images. [23:19:48] Why aren't we using SVG for the site logo on Wikipedia? [23:19:50] and [23:20:08] Why are the bullets in Vector images and not characters? [23:20:47] 1) The SVG is too heavy to be used as a logo. Not sure if IE supports SVG at all without add-ons. [23:21:05] 2) No idea, but MatmaRex will know :-) [23:21:22] StevenW: because they are list-style-images [23:21:23] 1) also, SVGs are slow to render on low-end systems [23:21:36] StevenW: and there's no way to make a PNG fallback for that [23:21:52] Gotcha [23:21:54] ah, you said 'characters' [23:21:54] Thanks guys [23:22:18] then because there's no CSS property to set any characters as a list "bullet" [23:22:34] or at least none i'm aware of [23:22:53] StevenW, yah, browser support somethingsomething. Ping kaldari maybe? (if he's not already buried under that bug's emails) [23:23:11] ul { list-style-type: "★"; } [23:23:57] twkozlowski, Wikipedia is Communism? [23:24:09] ul li:before { content: "+"; } [23:24:22] StevenW: proper cross-browser support for SVGs just landed in core a few weeks ago, but we should be able to fix the logo now. [23:24:38] How? [23:24:53] MaxSem: you can use • inside your CSS :) [23:25:18] MatmaRex: how about those two ways? :) [23:25:24] ah yeah, what twkozlowski said is close [23:25:26] StevenW: basically, it's a clever shim/hack for PNG fallback on the proper browsers. [23:25:29] there used to be a display: marker, too [23:25:38] to make it display the way list bullets do [23:25:44] but it got offed in CSS 2.1 or 3, i think [23:26:07] does list-style-type accept arbitrary values? [23:26:31] http://dev.w3.org/csswg/css-lists/#marker-content [23:27:13] oh, i didn't know about ::marker [23:27:24] (scroll down to example 3) [23:28:32] twkozlowski: does anything support that? [23:28:34] kaldari: how do you imagine fixing the logo if what MaxSem says is true? [23:28:50] MatmaRex: no idea, but if W3 suggest that... [23:29:03] that's the spec [23:29:06] doesn't mean it works [23:29:08] kaldari: just curious :) [23:29:17] If Wikipedia is Communism, I guess there's no hope [23:29:46] <^d> kaldari: Least efficient communists ever ;-) [23:29:56] (I still think we oughtta re-write the Wikipedia globe before we ever serve it to users. 250+ KB of code is way too much.) [23:30:02] <^d> kaldari: c.f. [[WP:DEADLINE]] [23:30:23] * MatmaRex tests [23:30:51] twkozlowski: no, nothing does. [23:30:59] twkozlowski, StevenW: Ah, I didn't realize the SVG version was so huge [23:31:00] my test case was: [23:31:01] [23:31:01]

1
2
3
4 [23:31:22] probably better to leave it how it is for now [23:33:31] MatmaRex: the before: one might work? [23:33:42] ul li:before { content: "+"; } [23:33:51] + add some padding if you want to [23:34:09] twkozlowski: it could be made to work if we try really hard [23:34:25] StevenW is the curious one :) [23:34:56] with some padding, removing the regular one, user-select: none, some fallback for crappy browsers [23:35:15] StevenW: something for someone with a few hours to spare to investigate ;)