[00:24:37] looks like gluster 3.4 and 3.3.2 were released a couple days ago [00:38:30] god damn it [00:38:35] gluster seems to have a gid limit [00:38:43] guess who's over the limit? :) [00:39:40] 3.4 doesn't have this issue [00:39:54] also, the nfs server is giving me shit on tools-login [00:40:17] Coren: ^^ [00:40:52] Ryan_Lane: Yeah, it wedged again. It should clear up in a 30-60 secs. [00:41:02] in dmesg: nfsd: peername failed (err 107)! [00:41:19] Googling further, this seems to be a known issue with ESX since Nov 2012 [00:41:26] ESX? [00:41:29] nfsd peername == symptom [00:41:41] Yeah, they gots a linux kernel underneath, remember? :-) [00:42:14] ah. right. ESXi uses the bsd kernel [00:43:41] Hm. From what I can tell, it's about the kernel never getting interupts for completed commands under some circumstances; so the driver times out, reaps manually, and soft resets. [00:44:08] (Speaking of, the resets took place, NFS should be back) [00:44:25] it is [00:44:48] http://ganglia.wikimedia.org/latest/graph.php?r=day&z=xlarge&h=labstore3.pmtpa.wmnet&m=cpu_report&s=descending&mc=2&g=cpu_report&c=Labs+NFS+cluster+pmtpa [00:44:56] Orange spikes == wedged. [00:45:44] It was quiet during the (UTC-4) day; and has picked up a few hours ago. Definitely driven by traffic. [00:46:59] (The small spikes over the blue bumps aren't the controller being wedged -- they are puppet runs) [00:52:45] Ryan_Lane: According to MegaCli, everything is full of joy on the hardware side. [00:52:55] heh [00:53:18] maybe we should go with a stable kernel and drop the thin provisioning? [00:53:40] or is it already too late for that? [00:55:15] Ryan_Lane: It's doable, but would require dump/restore. [00:55:42] It'd be one hell of a regression though if it affects 3.5 /and/ 3.8 kernels. [00:56:15] I'm thinking something else might trigger the issue. Ima do some comparison with labstore1 [00:56:15] * Ryan_Lane nods [00:56:29] Has an H800 too, right? [00:56:34] this didn't happen in 3.5 did it? [00:56:39] * Coren nods. [00:56:50] But it was completely overshadowed by the 14-day problem. [00:57:07] it doesn't look like the same issue... [00:57:12] * Coren checks the older syslogs [00:57:18] I checked the older syslogs; there were resets in there too. [00:57:20] this completely blocks filesystem actions [00:57:39] Yeah; different problems though I wouldn't be stunned if one precepitated the other. [00:57:55] and yeah, they both have the same controller [00:58:23] There is also the dumb possibility of an actual hardware problem with the controller or the shelf. :-) [01:03:29] dear gluster, I fucking hate you [01:03:38] it's such a gigantic piece of shit [01:04:19] tried to delete and recreate a volume [01:04:23] god help you if you try that [01:13:08] Aha! [01:13:21] http://forums.freenas.org/threads/lsi-megaraid-sas-9261-8i-timeout-with-write-back.5960/ [01:13:36] labstore3 is in write-back; labstore1 is in write-through [01:14:25] Wrong OS, but same basic hardware. [01:16:41] Thankfully, that's easy to test: I just switched all devices to WriteThrough. [01:17:26] Ryan_Lane: ^^ [01:17:31] Now I sit back and wait. [01:18:18] heh [01:18:49] WMF ENGINEER ACTS RECKLESS WITH HARDWARE SETTINGS! LET US FORK ENGLISH WIKI!!!!1 [01:18:50] :) [01:19:19] well, this would be a more conservative setting ;) [01:19:35] with worse write performance [01:19:56] who cares?! I wasn't informed! [01:20:47] :) [01:22:11] Holy... [01:22:17] You have GOT to be shitting me. [01:22:30] I really don't understand why this one project is giving me issues with gluster and no other ones are [01:22:35] Coren: ? [01:22:36] http://ganglia.wikimedia.org/latest/graph.php?r=hour&z=xlarge&h=labstore3.pmtpa.wmnet&m=cpu_report&s=descending&mc=2&g=cpu_report&c=Labs+NFS+cluster+pmtpa [01:22:46] Look at the bunch of tiny spikes every 30s [01:22:57] My flush_dirty interval is set at 30s. [01:23:18] So now it does a bit of iowait when buffers are flushed. What you'd expect with writethrough right? [01:23:53] ... what if the timeout was caused by the controller doing all of /its/ writeback flush at once when its own cache got full? [01:24:17] That'd explain the decreasing interval as write traffic increases. [01:24:39] And, being over the time limit, would cause a soft reset (that'd politely wait until everything was flushed). [01:25:29] s/30s/60s/ [01:25:34] 600 centiseconds. [01:26:49] If setting writethrough fixes things, I'm going to be /very/ cross at whoever wrote that firmware. [01:27:19] YuviPanda: Where is that fork enwp thread anyways? I could use a laugh. [01:27:31] Coren: https://en.wikipedia.org/wiki/Wikipedia:Village_pump_%28proposals%29#Fork_the_wiki [01:27:43] Coren: When did you change the setting? Or does it need a restart? [01:27:52] scfc_de: I did, it doesn't. [01:27:54] Coren: they found our secret (WMF has other 'clients' we 'sell to', apparently) :) [01:28:28] scfc_de: If you look at the graph I pointed to, you can see the change of setting at the last blue spike after the big orange lump. [01:28:31] Is there anyone here that can restart Helpmebot by chance? retsreklawts seems to be asleep already for the night. [01:31:10] heh. seems I hit gluster's 64 gid limit [01:31:14] sad that I was in that many groups [01:32:16] in 3.4 the 64 gid limit is gone too [01:32:23] maybe I should upgrade right now [01:32:24] :D [01:32:29] * Ryan_Lane is kidding [01:32:35] Ryan_Lane: I'm pretty sure you can't make things worse. :-) [01:32:39] hahaha [01:32:57] gluster responding and serving filesystem is better than it shitting itself [01:33:02] * Coren would rather continue struggling with issues on an NFS server than attempt to wrestle gluster. :-) [01:33:20] I'm just going to wait until we replace it with NFS [01:34:18] But really, if that effing firmware implements writeback as "wait until my buffers are all full then flush them all at once", it wins the Gluster Prize of Excellence! :-) [01:34:45] * Coren decides 'Gluster Prize of Excellence' needs to be awarded to other truly horrid pieces of software. :-) [01:35:43] * YuviPanda awards Gluster Prize of Excellence to Mediawiki [01:35:58] Coren: The graph looks like a Rorschach test to me :-). The "Wait" line seems to be a bit more prominent after 1:15Z. [01:37:16] scfc_de: That's to be expected; setting the controller to writeback means that when the OS issues a write, it'll get the OK only once the block has actually landed on the rust; the orange spikes are the flush processes waiting on that IO [01:38:29] They used to be much flatter because then it only had to wait for the block to hit the controller's cache instead. [01:39:18] Coren: gluster would probably be better than NFS for us, if we didn't need multi-tenancy [01:39:31] alas [01:40:20] Ryan_Lane: I'm not giving up on the idea of a DFS, but we need to hammer it out seriously before we put stuff people expect to keep working on it. :-) [01:40:39] gluster fails because it can't handle the number of volumes we're running [01:41:05] can't do multi-tenancy without running multiple volumes [01:41:05] Coren: Really? I googled for H800, and for example the excerpt in http://en.community.dell.com/support-forums/servers/f/906/t/19398489.aspx defined write-back = "controller cache has received all the data in a transaction", write-through = "disk subsystem has received all the data in a transaction". "Disk subsystem" = superset? [01:42:10] OS buffers -> controller -> cache -> disk hardware -> disks [01:42:57] Coren: But then "setting the controller to writeback means that when the OS issues a write, it'll get the OK only once the block has actually landed on the rust" isn't true? [01:43:04] "disk subsystem", in that context, probably means the hardware that actually talks to the disks including the disks themselves. [01:44:11] That terminology is wonky though. It probably was written neutrally enough to avoid presuming that what's on the other end of the SAS cable is an actual disk. [01:46:16] Coren: So you meant to say "That's to be expected; setting the controller to *writethrough* means that when the OS issues a write, it'll get the OK only once the block has actually landed on the rust"? [01:47:00] ... I didn't notice that error, even though you pointed it out. :-P Yes, of course. [01:47:55] uuuuuggggghhhh. role::puppet::self regenerates the client key when you specify a central puppet master? [01:48:13] * Ryan_Lane stabs [01:48:16] * Ryan_Lane stabs hard [01:48:43] Coren: As long as the controller understands you correctly ... :-) [01:48:50] Good night everybody. [02:07:41] hm. these instructions for multi-node role:puppet:self doesn't work [02:09:10] *don't [04:08:59] [bz] (8NEW - created by: 2Ryan Lane, priority: 4Normal - 6normal) [Bug 51581] Deployment-prep deploys from master and uses a submodule with submodules - https://bugzilla.wikimedia.org/show_bug.cgi?id=51581 [05:15:18] Why tools-login is responding so slow? [05:16:06] because it's having nfs server issues [05:24:39] Ryan_Lane: again? [05:24:49] more like still [06:04:46] Ryan_Lane: you're probably aware en.wikipedia.beta.wmflabs.org is down again. Not a problem for me, just letting you know. It was working OK around 90 minutes ago. [06:05:04] spagewmf: I don't actually have anything to do with beta [06:06:22] Ryan_Lane: sorry, the .wmflabs.org in the hostname steered me wrong. I know hashar deals with it, should I let ops know? [06:06:40] hashar is the one that maintains it [06:06:50] k, thanks [06:06:51] no one in ops does as far as I know [06:07:43] I'm digging the gluster comments in the scrollback [06:14:18] Ryan_Lane: this is horribly slow [06:14:30] spagewmf: it's not glusrer [06:14:35] spagewmf: it's the nfs server [06:14:41] deployment-prep switched to nfs [06:23:10] Ryan_Lane: better now [06:24:09] zhuyifei1999: see: https://ganglia.wikimedia.org/latest/?r=hour&cs=&ce=&m=cpu_report&s=by+name&c=Labs+NFS+cluster+pmtpa&h=&host_regex=&max_graphs=0&tab=m&vn=&sh=1&z=small&hc=4 [06:26:23] Ryan_Lane: thanks [06:26:28] yw [07:18:32] Ryan_Lane: why did it happen? [07:23:32] Hi everyone. [07:23:38] For a little bit at least. [07:23:48] I'm in Germany now. [07:24:15] Cyberpower678: you ar in germany? who? [07:24:19] *hi [07:24:42] who? [07:24:46] Hi! [07:24:54] *where? [07:25:05] Gross-Gerau [07:26:01] And jetlagged. [07:26:20] * Cyberpower678 might be going to sleep in a moment to catch up on sleep. [07:27:24] Cyberpower678: Why Germany? [07:27:56] Because it is my birthplace and my nationality, my citizenship, and it's where my family lives. [07:28:32] I'm on a temporary mobile connection right now, so I will go offline later. [07:28:58] Just running some tests on my new script now. [07:31:52] :) [07:32:08] Steinsplitter, where are you? [07:33:28] North-Italy [07:33:51] Oh. [07:34:01] ca. 4/5 H from Muinch [07:34:26] I'm near Frankfurt [07:35:27] :) [07:36:41] I just landed. [07:41:37] I sincerely hope that's merely a spike on the load and not permanently going up again. [07:42:26] Yes it is. [07:49:59] . [08:10:56] zhuyifei1999, >:( [08:39:58] Coren: on toolabs the hu database is mostlkely missing revision_userindex [08:42:18] i just wrote https://wikitech.wikimedia.org/wiki/User:Legoktm/pywikibot_on_tools_lab, i think someone in here was asking help with it the other day [08:50:17] Coren: what do you think about https://bugzilla.wikimedia.org/show_bug.cgi?id=51310? [09:20:41] [bz] (8VERIFIED - created by: 2Aude, priority: 4Unprioritized - 6normal) [Bug 48743] wikidata_singlenode puppet manifest is broken - https://bugzilla.wikimedia.org/show_bug.cgi?id=48743 [11:38:23] T13|sleeps: ^^ [11:38:28] @ping [11:38:34] !ping [11:38:35] pong [11:58:50] !log bots installed nano on bots-labs :> [11:59:05] >.< [12:01:18] Logged the message, Master [12:01:52] Coren: did the person having SMTP issues eventually get it fixed? [12:01:56] forgot who that was :| [12:10:59] addshore why you install nano on bots-labs? [12:11:09] cause i like nano! [12:11:25] what happened to wm-bot [12:11:31] it died :> [12:11:38] howcome [12:11:55] unresponsiveness :D [12:12:04] you don't need to restart bouncers to fix that [12:12:27] hmm [12:13:47] that's weird [12:13:53] !pin [12:13:53] pong [12:14:11] petan: just scrolled up in my console [12:14:21] as far as I remember i didnt kill the bouncers :P [12:14:29] you didn't? [12:14:36] how did you restart the bot [12:14:42] *checks* im the wrong user actually [12:15:41] started by killing wmib.exe and re starting restart.sh [12:16:22] nothing happened :> so i killed the bouncers and restarted them also [12:19:07] petan: i also fixed your docs ;p https://wikitech.wikimedia.org/w/index.php?title=Nova_Resource:Bots/Documentation/wm-bot&diff=77756&oldid=77520 [12:24:13] Morning petan [12:24:25] addshore: https://wikitech.wikimedia.org/wiki/Nova_Resource:Bots/Documentation/wm-bot#How_to_fix_1_or_more_disconnected_instances_for_bot_which_is_running [12:25:05] :> [12:25:09] you should put that at the top ;p [12:25:24] BB [12:25:37] oh wait, thats new! :O [12:41:14] !log deployment-prep Text cache was not in wgSquidNoPurge, that caused all requests to be interpreted as coming from the text cache causing misc issue (such as throttling account creation for everyone). [12:41:17] Logged the message, Master [12:42:05] http://bots.wmflabs.org/~wm-bot/dump/%23wikimedia-labs.htm [12:43:52] !addshore [12:43:52] fail [12:44:01] lol ha.. [12:44:02] :< [12:44:22] I didn't do it, but find it funny.. [12:44:45] hehe i know :P [12:44:45] !accountreq [12:44:45] in case you want to have an account on labs please read here: https://labsconsole.wikimedia.org/wiki/Help:Access#Access_FAQ [12:46:22] !Coren [12:46:22] Coren is dead. petan killed him. He now roams about as a zombie. [12:46:32] !Cyberpower678 [12:46:33] addshore, how do you rollback? with the rollback button? :D [12:46:42] !demon [12:46:42] <^demon> Docs exist solely for developers to go "omg you didn't read the docs!" when people ask common questions. In practice, nobody reads docs before asking questions. [12:46:57] So true!!!! [12:47:35] !Cyberpower678 [12:47:36] addshore, how do you rollback? with the rollback button? :D [12:48:26] :> [12:50:33] T13|needsCoffee: I don't think this channel is for playing with bots [12:52:09] zhuyifei1999: just reading the stuff already there.. :) [12:53:03] T13|needsCoffee: #wikimedia-labs-offtopic [12:59:53] @trusted [12:59:53] I trust: petan!.*@wikimedia/Petrb (2admin), .*@wikimedia/.* (2trusted), .*@mediawiki/.* (2trusted), .*@wikimedia/Ryan-lane (2admin), .*@wikipedia/.* (2trusted), .*@nightshade.toolserver.org (2trusted), .*@wikimedia/Krinkle (2admin), .*@[Ww]ikimedia/.* (2trusted), .*@wikipedia/Cyberpower678 (2admin), .*@wirenat2\.strw\.leidenuniv\.nl (2trusted), .*@unaffiliated/valhallasw (2trusted), .*@mediawiki/yuvipanda (2admin), [13:08:52] meh [13:09:19] @channels [13:09:36] !ping [13:09:36] pong [13:09:45] zhuyifei1999, that's right. So don't make @kick you. :p [13:09:50] I am in 92 channels in this moment [13:10:02] wm-bot sure get around. [13:10:23] petan: is that with both nicks? [13:10:38] http://bots.wmflabs.org/~petrb/db/systemdata.htm [13:10:38] Cyberpower678: ? [13:10:52] wm-bot Online in 50 channels 6667 [13:10:53] wm-bot2 Online in 42 channels 6668 [13:11:05] @trusted always pings me zhuyifei1999. I'm an admin in this channel. [13:11:05] I trust: petan!.*@wikimedia/Petrb (2admin), .*@wikimedia/.* (2trusted), .*@mediawiki/.* (2trusted), .*@wikimedia/Ryan-lane (2admin), .*@wikipedia/.* (2trusted), .*@nightshade.toolserver.org (2trusted), .*@wikimedia/Krinkle (2admin), .*@[Ww]ikimedia/.* (2trusted), .*@wikipedia/Cyberpower678 (2admin), .*@wirenat2\.strw\.leidenuniv\.nl (2trusted), .*@unaffiliated/valhallasw (2trusted), .*@mediawiki/yuvipanda (2admin), [13:11:09] oops [13:11:10] OMG OMG [13:11:13] can you just stop using it [13:11:40] @trustdel petan!.*@wikimedia/Petrb [13:11:40] User was deleted from access list [13:11:41] petan, sorry. :p [13:11:46] here we go [13:12:13] Cyberpower678: petan is root [13:12:23] so don't worry [13:12:26] actually I am Peter [13:12:28] :3 [13:12:38] I am root and addshore is stump [13:12:51] zhuyifei1999, I know. [13:13:14] petan, and I'm the branches. [13:13:17] Cyberpower678: What's @trusted always pings me zhuyifei1999. I'm an admin in this channel. [13:13:20] :3 [13:13:48] zhuyifei1999, ?? [13:14:02] petan: what ever, but it's your nick [13:14:03] zhuyifei1999, when you use @ trusted. You ping me everytime, [13:14:21] Cyberpower678: ok [13:16:11] I am stump! [13:16:29] :P [13:16:36] addshore: you're trusted [13:16:39] better than root [13:16:46] at least cute chicks can sit on you [13:16:53] O_o [13:16:55] xD [13:16:55] petan: I thought you were Petr? :p [13:16:55] XD [13:17:03] Peter == Petr [13:17:06] how much beer have you had today petan ? :P [13:17:11] none yet [13:17:15] I am in office :P [13:17:19] Yet... [13:17:19] not that I couldn't have it [13:17:25] thats no reason not to have beer ;p [13:17:25] but I prefer coke in work [13:17:44] i drink so much water at work xD [13:17:54] This is seriously offtopic [13:18:14] @kick petan stop posting this off topic shit plz [13:18:14] !!log [13:18:17] petan needs a new hobby :P [13:18:24] :< [13:19:02] What timing... [13:19:40] petan: wm-bot kicked you then said you need a new hobby.. :p [13:19:40] petan: [13:19:42] (13:18:15) ChanServ!ChanServ@services. changed mode +o wm-bot [13:19:43] (13:18:16) User wm-bot!~wm-bot@wikimedia/bot/wm-bot kicked petan from channel with reason: stop posting this off topic shit plz [13:19:44] (13:18:17) petan needs a new hobby :P [13:19:46] (13:18:23) :< [13:19:47] (13:18:39) petan!~pidgeon@wikimedia/Petrb just joined the channel [13:20:04] :> [13:20:08] Priceless.. [13:21:29] addshore how big is your flat :D [13:21:35] not very big [13:21:37] october is sooon [13:21:39] XD [13:21:42] but i dont seem to spend much time there xD [13:21:48] you better make some party there lol [13:21:56] haha, i dont think thats possible :P [13:21:59] :( [13:22:08] maybe party somewhere else and collapse on the floor at the flat? :P [13:22:24] !addshore [13:22:25] fail [13:22:25] yes that work :P [13:22:29] lol [13:23:04] addshore you got many friends in there already? how are these WMDE folks [13:23:24] they are lovely :) [13:23:34] this reminds me I wanted to take andre klapper to some pub :D [13:23:37] he lives in prague lol [13:25:19] !addshore | addshore [13:25:19] addshore: addshore is no longer fail! [13:25:31] :O [13:25:43] what a trick [13:25:48] * prick :D [13:25:51] :D [13:26:00] hush now :) [13:28:56] !zhuyifei1999 [13:28:56] 9991iefiyuhz [13:30:09] !delete is petan deleted me once, and then andrewbogott came and deleted whole my server by accident. I tell you people, deleting software is evil and should be illegal. If you don't like some program, don't delete it, just shoot yourself or something... [13:30:09] Key was added [13:30:22] !delete [13:30:22] petan deleted me once, and then andrewbogott came and deleted whole my server by accident. I tell you people, deleting software is evil and should be illegal. If you don't like some program, don't delete it, just shoot yourself or something... [13:42:36] That server is cursed! Cursed I tell you! [13:50:02] nfs? :P [13:50:27] ok, gluster suck, nfs suck, is there anytthing else we can switch to [13:50:50] * petan quiets YuviPanda before he starts about SMB-crap [13:51:00] we could use SMB [13:51:04] :P [13:51:07] ... [13:52:57] Well, NFS doesn't suck. In fact, it's been incredibly robust given the fact that the @&#^ controller flakes out every 20 minutes. [13:53:46] And, BTW, it's called CIFS; SMB is the predecessor that sucked even more but that nobody has used for ~10 years. :-) [13:54:04] YuviPanda: What's SMB? Some Magic Bean? [13:54:14] LOL [13:54:21] !smb is YuviPanda: What's SMB? Some Magic Bean? [13:54:21] Key was added [13:54:38] Coren: more people know 'SMB', hence it works better :) [13:54:47] and I am pretty sure everyone agress that anything is better than SMB [13:54:51] here he is. the gut with a magic bean... [13:54:55] * guy [13:55:28] there is some fuckshitcrap [13:55:56] I am trying to find out how to call it [13:56:12] Coren forbid me saying "there is load over 20" so I will just say there is fuckshitcrap on grid [13:56:43] hm... but it's pretty constant [13:56:48] maybe nfs is not borked this time [13:57:05] nah [13:57:17] <^demon> Friendly reminder: Gerrit is coming down in a few minutes for some hardware work. Please don't panic. [13:57:30] NOOOOOOOOOOOOOOOO what are we going to do eeeeeeek [13:57:38] <^demon> Go outside and get some fresh air :D [14:01:18] What is an "outside"? [14:01:56] <^demon> Good question. I've only heard tales... [14:03:27] I was out once and I can tell you it's fucking scary [14:04:30] [10:04] DEBUG Exception in module Feed: The remote server returned an error: (502) Bad Gateway. last input was petan chan: #wikimedia-labs I was out once and I can tell you it's fucking scary [14:04:38] ... [14:04:41] I saw that [14:06:23] <^demon> petan: Your profanity scared gerrit-wm away! ;-) [14:08:12] ǃlog [14:09:33] !hashar [14:09:33] [10:15:12] !log WMFLabs seems to have recovered now [14:10:00] ǃhashar [14:10:23] !log ? [14:10:23] Message missing. Nothing logged. [14:10:28] <^demon> !me [14:10:29] bleh [14:10:38] !^demon [14:10:57] !demon [14:10:57] <^demon> Docs exist solely for developers to go "omg you didn't read the docs!" when people ask common questions. In practice, nobody reads docs before asking questions. [14:11:13] ^^^ True story... [14:12:24] <^demon> Technical_13: This is why FAQ stands for "asked" questions and not "answered" questions ;-) [14:12:49] ǃlog [14:12:54] I think I am blacklisted by the bot [14:13:02] must have been throttled [14:26:05] [bz] (8NEW - created by: 2spage, priority: 4Unprioritized - 6normal) [Bug 51580] configure beta labs for SUL2 - https://bugzilla.wikimedia.org/show_bug.cgi?id=51580 [14:32:52] <^demon> Ok, gerrit's back up. Things might be a little slow for a bit while the disks finish sync'ing, but nothing to worry about. [14:42:23] um... Coren: an issue with fcgi again [14:42:49] JohannesK_WMDE: please to give details. [14:43:13] typing! :) i checked whether the request function is actually called without the interpreter on subsequent request, and it isn't. [14:43:45] JohannesK_WMDE: did you get your SMTP issue solved, btw? [14:43:48] the script is always executed from the start, like a normal CGI script. i'm using flup. anything special i need to do to activate fcgi? [14:43:49] I'm not sure I understand what you mean. [14:44:11] Actually, I'm sure I don't understand. :-) [14:44:11] Do we support fcgi at all? [14:44:11] YuviPanda: yes, it seems to work now. [14:44:13] I don't think so [14:44:43] No FCGI support atm. I'd have been done with WSGI by now if it hadn't been for the NFS server's controller giving me issues. [14:44:53] yeah, that'll explain JohannesK_WMDE's problem [14:44:57] :) [14:45:00] ohhhhhh. okay. well that explains it :p [14:45:16] :P [14:45:29] i faintly remember you saying something about fcgi partly working, Coren [14:45:31] FCGI is on the plan, after WSGI which is a more requested feature. :-) [14:45:59] JohannesK_WMDE: No, I said FCGI would be easier to implement in this setup than WSGI, but the latter is waited on by more people. :-) [14:47:26] Coren: can you have a look at https://bugzilla.wikimedia.org/show_bug.cgi?id=51310 and tell me what you think? [14:48:46] well, i want to have the requests served without an interpreter being started each time. that is possible with wsgi, right Coren? i don't care about the first letter really... [14:48:52] giftpflanze: I /can/ create a dedicated instance with its own queue, but why do you think it's going to be necessary? [14:49:13] JohannesK_WMDE: Yes, wSGI is pretty much "FCGI on steroids" [14:51:48] Coren: hm, let me first try implementing it with the infrastructure given [14:52:00] but the Thread package would be nice [14:52:05] so, how much work is it approximately to make wsgi work Coren? [14:52:45] JohannesK_WMDE: About a day's worth, but right now I'm delayed by the NFS hardware/driver issues. [14:53:07] giftpflanze: I see no reason why that wouldn't work; is there a missing package? [14:54:07] it is named tclthread [15:02:57] Coren: I tried resetting my password on en.wikipedia, put in my email, and it never got sent to me. Is this something you could investigate for me? [15:07:49] rachel99: Not unless I could positively establish your identity, sorry. [15:11:25] coren: I really want to make it be one of the accounts for my SUL, but it won't work for the en.wikipedia, as apparently I have the wrong password for it. What do you suggest I do to fix this? [15:12:04] rachel99: Do you currently own the SUL account? [15:12:09] yes, I do [15:12:41] rachel99: Then you need not worry, when SUL finalization takes place, the enwp account will be renamed out of the way. [15:13:28]