[12:45:51] hannes-_-: do you have the URL for the remote study? Can you email it to me? [12:56:08] hi nimish [12:56:10] sue [12:56:13] sure [12:58:45] though thats for the test run - so i am not sure if the kink is still valid [12:58:52] the link [12:59:11] sry...typping with one hand - baby in the other hand [12:59:40] aww, hehe [13:00:13] yeah I don't think it is...I didn't get the URL from Paru for today, the first participant starts...now and she's not up yet (it's 6am our time) [13:01:20] oh it starts now? [13:01:52] yep [13:01:58] 9am eastern [13:02:17] we just found that out yesterday evening [13:02:35] hence me being up this early [13:06:37] so did you find out how we can join? [13:07:22] it doesn t work with the gotomedia quickconnect software [13:08:02] I am sure Parul oversleeps [13:11:26] I'm trying to contact the agency [13:12:27] wait [13:12:29] I ve got it [13:12:34] I am in [13:12:38] nimish_g: wait [13:13:51] *guillom waves [13:13:58] HI guillom [13:14:09] can u see nimish? [13:14:13] everything ok with the remote study? [13:14:18] please tell him to come to irc [13:14:19] no I'm still at home [13:14:22] oh [13:14:22] ok [13:14:30] hannes-_- what's the URL? [13:14:40] rather, can you email it to me? [13:15:15] guillom, I haven't been able to connect to it yet [13:15:33] sent on mail nimish_g [13:17:28] hey there u are? [13:17:32] ! [13:17:48] yes I am =) [13:18:25] guillom do you remember what the etherpad server is? [13:20:25] see pm [13:21:27] hmm, am I still on? My connection suddenly drops sometimes [13:22:29] you are :P [13:22:43] nimish_g: he doesn t use the prototype - is that correct? [13:24:06] he doesn't have FF [13:24:08] *guillom is going to do some sport and have breakfast now, seeya later [13:26:38] he could close some sidebars [13:27:11] that s ridiculous [13:27:25] and he could close some apps runing in the background [14:18:01] nimish_g: I would say we should not test on IE again [14:18:29] it makes more sense to let the prband to install FF first - even if that takes 5-10 minutes [14:18:36] what dou you think about that? [14:21:20] hi parutron [14:53:31] hey parutron [14:53:51] hey [14:54:14] how are u? [14:54:23] how is it going? [14:59:04] busy I guess [14:59:22] *guillom → office [16:11:57] the sound is terrible, I hope it's better on their end [16:18:07] are u wathcing in the conference room at the office? [16:18:38] yes [16:18:41] after much trouble [16:19:29] guillom I have an etherpad for this one too: http://eiximenis.wikimedia.org:9000/UXStudyAprilEarly [16:20:07] nimish_g, good, I was about to ask :à [16:20:09] :) [16:20:13] thanks [16:39:29] I can't understand a thing, I'm just going to go back to my desk [16:41:49] it works pretty well over here [16:41:54] sound is awesome [16:42:12] you should call on on your machine [16:50:04] glad to hear others' sound is great [16:50:06] and the recording [16:50:09] ours is, er, less than ideal [16:55:53] did you guys loose connexion too? [17:21:35] guillom [17:21:51] the first 2 edits made by the usability account on this article: http://en.wikipedia.org/w/index.php?title=Alex_Jones_%28radio_host%29&action=history [17:22:07] are legitimate and the rest aren't...do you know how I can roll it back to that version? [17:22:18] without clobbering all his edits? [17:22:39] nimish_g, go to the history, click the date & time of the good version [17:22:49] then edit this version and save without changing anything [17:22:54] you'll revert to this version [17:23:58] something is up with my tabs on Commons [17:24:10] It looks very wrong. [17:24:23] http://en.wikipedia.org/wiki/Help:Reverting#Manual_reverting [17:26:48] parutron, http://commons.wikimedia.org/wiki/File:Wikipedia%27s_W_%28Linux_Libertine%29.svg [17:27:31] SAY WHAT? [17:27:33] who made that? [17:27:48] I did. [17:27:56] YAY FOR CARY [17:28:00] OPEN SOURCE ALL THE WAY [17:28:02] tee hee [17:28:26] not sure what copywright/tm implications there are [17:28:30] Right [17:28:46] but i'm gonna throw them both into a file for side by side comparison! [17:28:54] cary = rockstar! [17:29:00] hannes_away: are you online? [17:29:10] cary's made a wikipedia W from linux libertine [17:29:21] btw - i owe you meeting minutes from our mtg yesterday [17:33:47] parutron: hats up? [17:34:09] *cary lifts his hat to hannes [17:34:47] thx cary ^^ [17:35:40] (standing ovation for hannes_away) [17:36:23] If that VV is not perfect, it's because it only took me 5 minutes to make. [17:37:15] havent seen it yet [17:37:21] but its a good idea [17:39:00] parutron, http://commons.wikimedia.org/wiki/File:Wikipedia%27s_W_%28Linux_Libertine%29.svg [17:39:16] heh... I didn't mean to poke Parul again [17:39:23] hannes_away, see link [17:41:40] sry have to be really away [17:41:47] will be back latewr [18:26:03] "Mincho" typefaces in Japan and "Ming" typefaces in China are basically the same thing. [18:26:32] *cary did http://commons.wikimedia.org/wiki/File:%E3%82%A6%E3%82%A3%E3%82%AD%E3%83%9A%E3%83%87%E3%82%A3%E3%82%A2_%28Wikipedia%29_in_Togoshi_Mincho.svg [18:33:28] hey cary, which app did u use? [18:33:30] inkscape? [18:33:37] Inkscape [18:33:44] Inkscape roxx [18:33:56] could u send me the original file? i d like to do some minor refinements ;) [18:35:00] That is the original file! :) [18:35:12] it's Inkscape [18:35:20] oh ok [18:35:22] It saves in SVG format [18:35:27] do you mind if I edit it? [18:35:42] I d like to correct the spacings at the top [18:35:42] I uploaded it as PD-ineligible :) [18:53:31] wow, the sound *is* really better [18:54:08] Is Roan around today? [18:54:47] I don't see him. [18:54:50] I don't think so; he said today and tomorrow were bad days for the usability study as far as he was concerned, so I guess he's busy elsewhere [18:55:00] Ryan_Lane, ^ [18:55:06] ah ok. good to know. thanks. [19:01:29] Did I mention that beta is all messed up on my Commons. [19:04:37] guillom: thank priyanka! [19:04:49] she installed a VM on her mac and hooked us up [19:04:57] i mean thank pdhanda [19:04:58] *guillom thanks pdhanda [19:08:56] np [19:14:10] It's RoanKattouw [19:14:14] Ryan_Lane|away was looking for you [19:14:30] Oh [19:14:44] I'll read the logs after I finish catching up on news and e-mail [19:32:53] parutron_: For your peace of mind, I'm here now [19:33:25] haha thanks [19:33:26] On a regular day I would probably have managed to be around from 6-7am PDT but I had to go to uni then to a birthday [19:33:27] nimish is here! [19:33:41] YAY for birthdays [19:33:47] fyi - it was trevor's bday yesterday [19:33:50] (in case you didn't know) [19:34:33] http://commons.wikimedia.org/wiki/File:%D0%92%D0%B8%D0%BA%D0%B8%D0%BF%D0%B5%D0%B4%D0%B8%D1%8F_%28Wikipedia%29_in_Russian_in_Linux_Libertine.svg [19:34:46] I knew that, yes [19:34:47] *cary is having fun now [19:37:13] cary: libertine rocks, doesn t it? [19:37:19] Indeed [19:37:32] I ve tested it in several languages [19:37:46] works pretty well [19:38:04] cary how do you feel about using a small last letter? [19:38:39] I don't like that idea. [19:39:04] do you think the capital last letter is important for WP recognition? [19:39:08] or why? [19:39:32] no one was on board with the small A [19:39:34] :( [19:39:38] except ME [19:39:43] in time [19:39:44] in time [19:40:23] the capital A is a typogrphical error [19:40:53] and I don t feel like it is important for the logo [19:40:58] the globe is strong enough [19:41:06] AND we got the very remarkable W [19:41:11] I think it's a consideration for a later date [19:41:22] yeah you might be right [19:41:41] actually I was sure that we won t change the last letter to be small [19:41:43] ;) [19:43:14] *hannes--- lifts his hat for cary and prepares for leaving [19:43:42] nimish_g_: does the etherpad save automatically? [19:43:51] it does [19:43:51] can I close it? [19:43:53] cool [19:43:55] yea [19:44:40] bye everyone and see u tommorw [20:10:44] guillom: if i were you, i'd consider having sean do your remote interviews [20:11:04] sean == much better moderator [20:11:16] will discuss with you when we're done here. [20:11:18] and you're saying that on a publicly logged channel :P [20:11:18] ok [20:11:37] eeps [20:11:40] but just for us [20:11:41] right? [20:11:42] well [20:11:48] i'll stand by my statement [20:11:50] ;) [20:12:12] héhé [20:25:49] RoanKattouw: Do you have some time to talk about monitoring? [20:25:53] Yes [20:26:01] *Ryan_Lane cheers [20:27:37] RoanKattouw: I'd like to monitor CPU and memory resources. Knowing how the resources are being used over time lets me know when I should add more resources to VMs. [20:27:57] RoanKattouw: Also, it would be nice to have alerts set for thresholds. [20:28:31] OK so it sounds like we should add the individual VMs to Ganglia at lesat [20:28:48] 75% (real) memory and 85% CPU would be likely be good at first [20:29:18] 85% CPU may actually be low. 95% might be better. [20:29:59] OK [20:30:14] So we want Nagios to throw an alert when mem/CPU go through the roof on any VM [20:30:19] Not sure Nagios checks that for us currently [20:30:36] Traditionally, Nagios checks are for stuff like is the host up (ping) and serving HTTP requests etc. [20:30:45] Although I have seen Nagios check for disk fullness [20:30:45] Yeah. Would be good. I can't really do much to add more CPU resources, but I can add more memory. [20:30:52] So I guess anything is possible [20:31:06] Storage space going above 80% would also be a good threshold for a warning [20:31:27] We have three disks on commons prototype, so 80% on each disk would be good [20:31:34] I think we have a cluster-wide threshold for that that's higher [20:31:41] ah ok [20:32:02] Most of the disk space warnings fire at 95% or 97% but then these are warnings on the order of "only 3% (350GB) left" [20:32:06] Whatever the cluster-wide threshold is, should be fine [20:32:11] ah [20:32:19] yeah, we have way less than that. [20:32:49] and having that low of disk space on / would be a fairly bad thing [20:33:16] Ryan_Lane: http://eiximenis.wikimedia.org:9000/swAfbC9sV6 [20:33:57] I can't access that from here :( I'll email myself the link [20:34:42] Oh dammit [20:34:46] It's to an Etherpad [20:35:00] So e-mailing yourself the link and reading it later is of limited usefulness [20:35:04] ah [20:35:25] they firewall pretty excessively here [20:36:10] OK so in addition to adding the VMs to Ganglia and adding Nagios alerts for CPU/mem/diskspace [20:36:16] So, these VMs are network segregated from the rest of the cluster. Will that be an issue for ganglia? [20:36:20] We'll want Nagios alerts for specific services [20:36:54] Ryan_Lane: Not necessarily. Ganglia divides boxes into groups, and each box multicasts its data to the entire group [20:37:06] Which is theoretically pretty fast if they're all on the same machine [20:37:19] Ok. [20:37:29] Then one node in each group is responsible for gathering the metrics and forwarding it up to the master gatherer [20:38:37] It may be good to set ganglia up for the selenium grid systems as well. Nagios isn't too important for them, as the grid server will show if they are available. [20:39:23] Ganglia and nagios for grid.tesla.usability.wikimedia.org is probably a good idea since it will act as a varnish server for the prototypes though. [20:39:47] Right [20:40:01] Well tesla itself should obviously be in Ganglia [20:40:11] But it'd be sweet to have the inidividual VMs in there as well [20:40:25] tesla can't, unfortunately [20:40:30] Right now all nodes in Ganglia are real machines, but we never used VMs in-cluster before so there's no precedent [20:40:40] Why is that? [20:40:58] What Nagios checks would you want for the grid server itself? [20:41:22] it is ESXi. Unless ganglia and nagios are going to get all information via SNMP, you can't get anything out of it. [20:41:43] it's an appliance [20:42:03] the grid server will be the reverse proxy for the prototypes, if it goes down, the prototypes do too [20:42:44] VMs are real systems from the POV of nagios and ganglia [20:42:48] OK so with individual HTTP checks on each VM serving HTTP reqs, we'd cover for reverse proxy failure already [20:42:51] Yes [20:43:21] Do you mean like there's no processes on tesla that can be run outside a VM? [20:43:27] *RoanKattouw has no idea what ESXi or SNMP are [20:43:31] ah [20:43:59] so, tesla is a virtualization host running ESXi. ESXi is an appliance OS that is essentially just a kernel [20:44:07] Right [20:44:11] So it can't run gmond [20:44:18] (Ganglia monitoring daemon) [20:44:23] yep [20:44:31] OK [20:44:34] SNMP is a network monitoring prototcol [20:44:43] remote systems can pull information using SNMP [20:44:56] most monitoring suites support it [20:44:58] So ganglia for that box is basically off the table unless we write a Ganglia/Nagios plugin for SNMP [20:45:03] Or maybe it's supported, no idea [20:45:09] I'd imagine they support it [20:45:20] Either way, I'm not sure how useful monitoring the box itself is when we're monitoring all its VMs [20:45:47] Well, it is actually useful to monitor the host as well as all the guests [20:45:48] We'd throw them into a group of VMs and that'd give us total CPU and mem numbers [20:46:19] If the host's memory or CPU is getting out of control it'll destroy the performance of all of the VMs [20:46:26] True [20:46:38] In ESX, it is normal practice to overallocate memory, CPU, and storage [20:46:45] Added that Ganglia to the list for tesla then [20:46:51] that's why I was tracking allocated storage using semantic mediawiki [20:47:08] Not sure what you'd have Nagios test there other than maybe CPU/mem as well [20:47:18] Disk storage should be fine provided we don't overallocate it [20:47:29] if you over allocate too much storage, then your VMs eat all of their space, they will all die :) [20:47:38] Yes [20:48:05] I've found everyone always over allocates space. it is really normal [20:48:10] Right [20:48:19] either way, I've been tracking it, so it shouldn't be a big deal [20:48:45] Disk space threshold can probably be a little higher on tesla itself [20:48:50] How large is its filesystem? [20:48:57] gimme a sec. lemme check [20:49:12] 1.36TB [20:49:18] If it's anywhere North of a couple hundred gigs something like 97% should be fine [20:49:19] Aah [20:49:30] 3% of that is some 31GB [20:49:37] that isn't high enough [20:50:05] Hm [20:50:07] 100-150GB would be better [20:50:18] if we are using that much space, we've almost definitely overallocated space [20:50:20] Considering that the VMs all die when the host runs out of disk space, I guess you're right [20:50:29] It's more severe than a simple disk full error [20:50:33] yeah :) [20:50:48] So let's say 90%, that'd start screaming at 136GB free [20:50:56] that would be good [20:51:28] OK [20:51:32] So lemme give you what I have now [20:51:37] Is pastebin.com firewalled too? [20:51:49] nope. it's actually the port that was a problem [20:52:18] Oh [20:52:36] OK I'll put it in an actual Etherpad then [20:52:43] ah ok. that'll work [20:53:19] http://etherpad.com/77X0SOq9wP [20:56:32] OK so for HTTP we have a protocol-specific check [20:56:49] It'll actually fire off a GET / and expect a 200 or 301 back (depending on which you set) [20:56:56] We'd want something similar for Selenium, I imagine [20:57:25] yeah. we can do a GET /console on 4444, and it should return a 200 [20:57:47] OK [20:57:48] Noted [20:59:12] Ryan_Lane: OK so does this list cover your wishes as to monitoring? [20:59:27] If so we should talk about notification next [20:59:30] RoanKattouw: We may also want to get network statistics as well. We had perfomance issues on the linodes due to network at one point [20:59:40] Ganglia collects those [20:59:46] ok. cool. [20:59:53] i'm good then [21:00:32] notification is an issue... I'm essentially in a communications black hole here [21:00:37] So, notification. If and when shit hits the fan, you and me should probably be notified by SMS or something [21:00:52] ah. I *can* get notified on google voice. [21:00:57] And an ops person I guess [21:01:03] Well surely you have a cellphone? [21:01:15] But GVoice is fine I guess? :) [21:01:30] I have a cellphone, but I can't bring it into work [21:01:33] Ah [21:01:37] GV number is: (504) 383-5263 [21:01:47] So the problem is the definition of "shit hits fan" [21:01:57] The ops department is still kinda struggling with that itself [21:02:02] last I heard [21:02:29] if you guys aren't able to work, or if your work is hindered because of performance issues, I consider that shit hitting the fan [21:02:33] But if we can come up with a reasonable threshold for when we'd like SMS and other notifications to be sent, Fred will probably like that [21:02:37] Right [21:03:19] I'm guessing that us being texted by a bot is probably not necessary in that case, as the person actually being hindered should notify us [21:03:27] if an alarm is being triggered, it can SMS me. [21:03:35] Right [21:03:41] GV SMS is free [21:03:56] I hope we can do something like "only SMS me if the box has been down for > 5 mins" because Nagios flaps sometimes [21:04:30] that sounds good. [21:07:44] OK [21:07:57] Then I guess we're pretty much done [21:08:07] I'll hash this out into a wiki page and poke Fred about it [21:08:15] ok. sounds good. [21:10:27] RoanKattouw: I should sign off. good talking to you. [21:10:31] OK [21:10:33] Thanks for being here [21:10:44] you're welcome [21:20:57] parutron, sent [21:21:45] cary: How was/is Beta all messed up on Commons? [21:22:00] I just sent parutron the screenshot [21:22:11] thx [21:23:12] RoanKattouw: (message from nimish_g ): can you push the code that you were supposed to push tomorrow? [21:23:24] Now? [21:23:36] yes, that's what nimish_g wanted me to relay to you (he's in a mtg) [21:23:40] OK [21:23:46] is there a reason not to? [21:24:15] No [21:24:41] cool [21:25:51] Any idea when Nimish will be out of that meeting? [21:27:17] Trying to reach him on Google Talk now but I'm not sure that'll work [21:27:20] Did he bring his laptop? [21:28:19] no [21:28:27] RoanKattouw: i saw nimish_g in the conference room interviewing the volunteer [21:28:27] he's interviewing my volunteer [21:28:31] Ah [21:28:33] OK [21:28:34] do you want me to get him? [21:28:45] i'm sure he can take a few minutes out [21:28:46] Please. I'll tell you what to ask, sec [21:29:12] Ask him whether I should revert his revert dated March 26 of the CSS changes [21:30:15] i think he is waiting on for you to deploy the bug fixes he requested yesterday [21:30:34] Yes that too [21:30:56] But additionally, I think he or Parul (or both) wants me to undo the undo of the CSS changes to balance the study [21:31:34] nimish_g will get back to you directly [21:31:39] he just grabbed his computer [21:31:46] OK [21:35:35] cary: I'm gonna hack around the issue you reported, it'll be resolved in a saner way eventually [21:35:43] \o/ [21:35:48] Basically I'm gonna assume that all JS-added tabs are collapsible [21:36:00] RoanKattouw, yes, please. [21:36:54] I'm sure people can have their own side tabs if they want to use beta *and* reduce steps. [21:37:11] Sure [21:37:17] cool [21:37:29] We could write a "move everything from the dropdown to proper tabs" Gadget, would be fairly trivial [21:37:54] what's going on in Commons? [21:37:59] hell. [21:38:13] Dante's inferno style. [21:38:26] Okay, I'm being dramatic. [21:38:27] i don't see anything odd [21:38:46] nkomura: All the JS-added tabs aren't getting collapsed [21:38:58] So if there's a lot of them and you have like an 11' laptop... [21:39:08] cary: are you in the office next thursday 4/8? [21:39:11] ah [21:39:17] parutron, I can be. [21:39:22] is it specific to commons? [21:39:23] *cary generally hates Thursdays in the office. [21:39:25] are you not normally? [21:39:27] :( [21:39:30] cary: Hack added, please Shift+Refresh and try again [21:39:34] would like to touch base with you and jay next week - was hoping for thursday am [21:39:50] is it because of the change we made to accommodate CatScan? [21:39:51] 10am, 11am, noon [21:40:05] nkomura: It's a bug in the modifications I made to move JS-added stuff from the dropdown to actual tabs on Commons [21:40:09] Yes [21:40:36] nopd [21:40:39] nope [21:40:41] RoanKattouw: whenever you're ready to push the CSS changes would be lovely [21:40:50] also, some of nimish's changes didn't make it into production [21:40:56] Still messed up. [21:41:00] am hoping those will get pushed too. [21:41:09] RoanKattouw: did you get my msgs abt that? [21:41:27] nimish_g: Which ones [21:41:40] Howie says I need to deploy your latest wmf-deployment rev [21:41:42] But [21:41:59] Do I also need to revert the rev where you revert some CSS changes because of study contamination? [21:42:49] haha yes. in the end, we want the templates to be black, no text decoration in underline [21:43:24] Just tell me which revs :) [21:43:36] I'm now in the process of deploying r64464 [21:43:49] And I believe I need to revert r64223 on wmf-deployment and deploy that [21:43:50] RoanKattouw, you got my comment above "nope" as in it's still broken? [21:43:53] Anything else you need? [21:44:01]