[00:59:27] Coren: Poke? [04:56:09] Coren: awesome, thanks :) [06:57:17] o_O Why global locked accounts can editing o_O [06:57:28] http://deployment.wikimedia.beta.wmflabs.org/w/index.php?title=Special:AbuseLog&wpSearchFilter=12 o_O [06:59:48] 01:29, 21 June 2013: User: (en.wikipedia.beta.wmflabs.org) triggered filter 12, performing the action "edit" on User:9h1k8do4b5. Actions taken: Disallow; Filter description: More Chinese spambots (details | examine | adjust visibility)' [06:59:49] And this seems to be a bus. "User:" o-O [07:04:06] *bug [10:07:41] Coren: is it on purpose that the superglobal $_SERVER does not contain "HTTPS"? [10:08:02] or petan maybe? [11:03:19] Kai_WMDE hi [11:03:27] hi petan [11:03:51] tbh I don't know :/ this is undocumented and it was set up by Coren [11:04:19] oh, i see [11:04:43] but thanks for answering :) [11:29:27] JohannesK_WMDE ping [11:29:34] JohannesK_WMDE you are running local-render ? [11:29:54] petan, yes we are, why? [11:30:07] there is instance of tlgbe/tlgwsgi.py on tools-webserver-01 [11:30:20] it eats 1600mb or resident memory [11:30:30] possible [11:30:34] 2gb of virtual memory [11:30:51] server has only 2gb of ram [11:30:54] that is a bit much but still possible [11:31:00] that's bad [11:31:13] is that normal? why it isn't optimized a little bit... better? :p [11:31:25] optimized? [11:31:36] yes, for example it could be split to multiple processes [11:31:44] or use some alternative storage than ram [11:31:51] why it eats so much operating memory? [11:32:04] so the 2GB of ram would be split into separate processes? what would that gain? [11:32:11] * Cyberpower678 is eating buttered petan. :D [11:32:12] we make huge db queries [11:32:17] it could be distributed to multiple servers [11:32:28] do we *have* multiple web servers? [11:32:31] yes [11:32:49] huge db queries do not necessarily need to use huge ram [11:32:52] i only know of 2. and i don't know of a way to distribute them. [11:33:34] you can't distribute 1 webserver session to multiple servers anyway... but I fail to see why it should eat so much, what kind of db queries it is? [11:34:05] if 1 web user is able to run a session which eats 1.6gb of ram, multiple web users could bring whole webserver cluster down [11:34:25] angelika fired off a query for just about every person in dewiki. that's probably what was so large. [11:34:53] you realize that running application which allows this can lead to abuses of servers? [11:35:19] imagine wikipedia would allow people to run expensive queries... like filter last 99999999999999999 contributions etc [11:35:28] it would bring whole wikipedia down in minutes [11:36:10] such an application if it's publicly accessible can be very dangerous [11:36:32] normally queries wouldn't be so large. but seriously, 2GB for a web server is too small. our smartphones have more. [11:36:44] 2gb for 1 session? [11:36:49] are you serious? o.o [11:36:54] JohannesK_WMDE: complicated processing should not be done on a web server in the first place ;-) [11:36:59] 2gb for the whole web server, petan! [11:37:11] I don't think production servers have significantly more [11:37:17] valhallasw: it's not complicated, it's just a large result set. [11:37:30] well, even ortelius on toolserver had 4 GB. [11:37:40] JohannesK_WMDE: as in: you're sending 2GB of data to the client? [11:38:41] valhallasw: i'm not sure. i am not allowed to log in to the webserver so i can't see what's happening. we never had 2 gb in one session allocated before. (btw: that's total apparently; rss is 160MB) [11:39:13] what you mean by rss [11:39:26] if you mean resident memory, it's 1540mb atm [11:40:05] petan: an. you said 1600MB rss, not 160. that's too much. still, i can't see what's happening. [11:40:45] how would you see it? [11:40:59] i can't because i can't log in to the webserver. [11:41:11] i asked coren yesterday if i could but he didn't reply. [11:41:13] it's a running process, you can't just see what's happening even if you had access to that server [11:41:29] ah. i can't but you can. ok. [11:41:57] I mean... how would you debug a running python process? [11:42:14] i could at least verify how much memory it uses, cpu time, etc. [11:42:56] it uses 3:43.56 cpu time 1546mb of resident memory and 2011mb of virtual memory [11:43:31] constantly on 25% cpu usage (1 cpu system) [11:45:43] petan: it's a 2GB per process limit, right? [11:45:48] or does the entire server have only 2G? [11:45:58] entire server has only 2g... [11:46:22] ... [11:46:22] ok [11:46:30] I guess the exec nodes have a lot more [11:46:55] petan: how much do the exec nodes have? [11:47:15] 8g [11:47:21] hmm [11:47:22] ok [11:50:09] petan: yes, this is a runaway thing, it is not intended to eat that much ram. we can do 2 things: first, restrict queries if we figure that the initial catgraph result set is too large. second, we can install a web server on sylvester and run the backend there. [11:50:34] petan: but you still want to put more ram into that webserver. 2gb for the entire tools webserver? you can't be serious. [11:51:08] I don't know why Coren decided to use only 2gb per webserver, but keep in mind there is going to be multiple webservers [11:52:09] petan: time to install slayerd ;-) [11:52:30] :/ [11:53:13] (which kills processes if a user uses more than a given amount of memory) [11:53:33] JohannesK_WMDE: is there any reason why catgraph cannot be submitted as SGE job? [11:53:52] or is it just very memory-intensive, while returning within ~10s for most queries? [11:54:25] yeah, I'm guessing Web Servers aren't supposed to do long running processes [11:54:25] valhallasw: it's an on-demand service [11:54:55] valhallasw: I suppose we don't really have proper precedents for doing this kinda 'background job' stuff. [11:55:32] Kai_WMDE: I understand that. But even then: you could send the user to a static html page that refreshes every 30-or-so seconds, which will be overwritten by the result from the query. [11:56:01] valhallasw: no, you misunderstand. catgraph isn't running on the webserver. [11:56:02] or use something like http://www.celeryproject.org/ [11:57:09] YuviPanda: I think the main problem lies in communicating to the user 'we are working on your job!', and then refreshing once the result is there [11:57:18] but this sounds like a generic problem [11:57:26] valhallasw: IIRC you can handle that with a little bit of ajax. [11:57:38] valhallasw: indeed. [11:57:59] addshore and I are working on a dumpscanner that would need something like this, too. [11:58:17] celery + redis sounds like a very nice solution to this [11:58:21] that would need something like? :D [11:58:23] and I think I'd want in to help :D link? [11:58:26] * addshore looks up [11:58:43] YuviPanda: http://tools.wmflabs.org/dumpscan/ [11:58:48] not working yet, though ;-) [11:58:53] its near working :P [11:59:01] YuviPanda: want to make it look pretty ? :P [11:59:02] but we are using a poor-mans alternative to celery + redis: the filesystem ;-) [11:59:10] so [11:59:18] it lets you scan the dump for files that match a specific regex? [11:59:26] and you're grepping through the logs everytime people run this [11:59:44] valhallasw: and no, catgraph can't be submitted as an SGE job, in the same way that, say, apache can't be submitted as an SGE job [11:59:46] addshore: i'm no designer, but can throw bootstrap on top of it pretty easily [11:59:52] hehe ;p [11:59:54] cu later, eat [12:00:09] JohannesK_WMDE: hmm, perhaps it can then run on its own labs instance? should be pretty easy to pick that up that way [12:00:58] addshore: valhallasw is this python or php? [12:00:59] YuviPanda: added you to the project [12:01:08] php frontend, python backend [12:01:08] hmm [12:01:14] YuviPanda: the html pages it html and a bit of php [12:01:17] backend is python [12:01:24] is it version controlled? :D [12:01:25] * YuviPanda logs in [12:01:34] also: https://github.com/addshore/dumpscan [12:01:47] i'm intereted in setting up celery + redis for something, so this might be a good candiate [12:01:49] * YuviPanda checks [12:02:25] basically, the job is now stored as json file and php (in the future) will submit an sge job to parse it [12:02:28] I imagine it could be incorperated into the labs infrastructure [12:02:33] yeah, I think so too [12:02:47] valhallasw: addshore mind if I convert the PHP script to python? WIll send a pull request... [12:03:00] YuviPanda, go for it :P [12:03:04] sweet [12:03:09] * YuviPanda uses Flask [12:03:16] it is super effective [12:03:18] * addshore really needs to step up his python ness [12:03:22] flask? [12:03:35] yes, frees you the pain of doing CGI by hand [12:05:50] addshore: http://flask.pocoo.org/ [12:06:22] :> [12:09:37] addshore: I'm just going to replace the current php code with python (writing it now) [12:09:45] addshore: soon we can move it to celery, etc. [12:09:48] this is a nice use case! [12:12:00] :P [12:13:58] * addshore needs to find a way to scan wikidata dumps easily :D [12:17:52] addshore: one of the things we can do later on is to invent nice readonly fast indexing of the dumps [12:17:57] on something that isn't mysql [12:18:04] addshore: labs is a lot of fun :D [12:18:08] YuviPanda: yes, i was also thinking that [12:18:12] and YuviPanda yes :) [12:18:21] addshore: we can also do something like have a neo4j instance [12:18:22] it has so much potential and easy ways to expand in all directions ;p [12:18:24] with all the graphs [12:18:29] and then we can do fast queries [12:18:34] that's one of the things that bugged me about ts [12:18:39] it was... technically limited. [12:21:59] mysql query can be run using sge? does it make sense? [12:22:38] it doesn't really need to. [12:23:54] YuviPanda: because are runned on different servers? [12:24:02] yeah, they're run on the db servers [12:24:08] the client is doing negligible work [12:24:52] YuviPanda: so I can even run them from webserver without any problem, isn't it? [12:25:06] should be, but why not just run it from tools-login? [12:25:11] assuming you are talking about the command line client [12:25:22] and not a separate program [12:25:48] YuviPanda: I'm writing a program that runs query (on crontab) [12:25:58] ah, then that should be on SGE fale [12:26:12] because programs take up memory, CPU on both the machine they are run and also on the mysql stuff [12:26:30] YuviPanda: the only thing the crontabbed program does is running query, nothing more ;) [12:26:41] what language is it in? [12:26:42] python? [12:26:44] php [12:26:53] so loading the php interpreter, etc takes memory :D [12:26:59] :D [12:27:03] fale: make the crontab start the job on SGE, shoudl be simple enough no [12:27:37] YuviPanda: yeah :) I used the same technique for a different program on TS :) [12:27:43] :) [12:28:01] sge has full db and (/home) fs capability? [12:28:35] yes [12:28:38] well [12:28:40] not /home [12:28:44] it has /data/project [12:28:48] and you shoudl be running it with a tool account [12:29:04] fale: after reading up, yes just do it on whatever server your on (so im guesing your suerying from a webserver / webpage) :) [12:29:08] *querying [12:29:25] YuviPanda: I do have a tool account, but I thought that tool account have the home folder in /home :D [12:29:44] fale: tools account home is in /data/project/toolname [12:29:49] :) [12:29:52] addshore: :D thanks [12:29:54] yeaah, do a pwd [12:30:07] addshore: even with crontab stuff? [12:30:22] yup [12:30:27] cd ~ [12:30:57] addshore: I was speaking about your phrase "yes just do it on whatever server your on (so im guesing your suerying from a webserver / webpage) :)" [12:31:10] :> [12:31:59] better keep the query synced to use less memory or to async them (and parallelized) to use resources for less time? [12:32:45] legoktm: do you remember which project we setup flask for? [12:33:42] addshore: i've written it, testing now [12:37:34] petan: around? [12:37:40] sure [12:39:13] petan: can you grep for errors from dumpscan.py? [12:39:22] in apache logs? [12:39:22] yes [12:40:22] petan: okay, just the last couple of exceptions? [12:42:26] YuviPanda: pastebin.com/vSWniRXU [12:43:03] errm, permission denied? [12:43:06] * YuviPanda scratches head [12:43:12] hold on o.O [12:43:25] try now [12:43:29] it wanted captcha :D [12:43:43] petan: no, i gave it captcha :) [12:43:48] petan: the error itself is confusing me [12:43:52] ah ok [12:47:31] petan: there should be a new error now, can you paste me it? [12:47:39] all this should hopefully go away once we have wsgi [12:48:15] no new error :( [12:48:27] maybe try to check your php error log file [12:48:47] this error logfile show only errors which were not catched by other logger [12:48:50] this is python [12:49:04] last error I have is from 12:37:16 [12:49:13] what is the value of date? [12:49:23] ? [12:49:37] wait I start irc client there [12:50:03] hmm? [12:50:21] I am on many computers now [12:50:25] (physical) [12:50:33] ah [12:50:34] ok [12:50:35] * YuviPanda waits [12:50:44] ok here [12:51:00] [Fri Jun 21 12:37:16 2013] [error] [client 10.4.1.89] Premature end of script headers: dumpscan.py [12:51:06] this is latest [12:52:22] are you sure? [12:52:26] http://tools.wmflabs.org/dumpscan/cgi-bin/dumpscan.py/test is giving me a 500 now [12:52:34] now not [12:52:42] [Fri Jun 21 12:49:53 2013] [error] [client 10.4.1.89] File "/data/project/dumpscan/cgi-bin/dumpscan.py", line 7, in [12:52:43] [Fri Jun 21 12:49:53 2013] [error] [client 10.4.1.89] /data/project/dumpscan/cgi-bin/app.py [12:52:45] [Fri Jun 21 12:49:53 2013] [error] [client 10.4.1.89] Premature end of script headers: dumpscan.py [12:52:46] 3 more [12:52:54] maybe apache is writing with delay [12:53:20] is that all it is giving you? no traceback, anything? [12:57:05] this is all I see in log :/ [12:59:19] fucking CGI [13:02:12] :D [13:03:54] err don't break the friendly channel policy :-) [13:04:19] it'll be easier to not do that once we have WSGI on :) [13:04:28] addshore: so it is currently writing folders into ~/scans [13:04:46] addshore: let me commit it and send a pull request :) [13:04:59] YuviPanda: mehehee [13:05:11] link me to it when your done :) [13:05:18] addshore: it is also using redis to find next-id [13:05:22] no race conditions :) [13:05:28] xd [13:05:35] addshore: it's already in the tool, checkout cgi-bin [13:05:45] mhhhm, cba to login atm ;p [13:05:45] am developing there, then copying to local for committing :) [13:05:47] h all [13:05:48] wiritng something ;p [13:05:49] *hi all [13:05:52] hi nutz_ [13:05:54] cba? [13:05:56] ah [13:05:56] ok [13:07:29] i'm on tool labs right now and am wondering: is it okay if i created a table which will probably have a bunch of GB? namely i want to run a CREATE TABLE x_revision AS (SELECT * FROM revision WHERE ... ); [13:10:02] in the documentation (https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Creating_databases) there is no limit written down, but i thought i'd better ask first [13:12:18] petan: Coren ^ [13:13:19] nutz: We're on a "as reasonable" basis atm with no hard limits, but the example you give doesn't seem like the best idea. [13:13:32] nutz: (For one, you won't get the SSD performance that way) [13:13:51] nutz: Why do you want to make a copy of revision? [13:17:01] Coren can you have a look why nfs is borked on toolsbeta [13:17:03] Coren: i have a bunch of requests that all query the same subset of revision with rev_user_text in (.... ) [13:17:29] and i figured that this way the queries would be faster since i dont have to run it through the whole revision all the time [13:17:47] nutz: Given the SSD and indexes, you're almost certainly better off trusting the optimizer [13:17:56] petan: Did you see my comment on your bug? [13:18:03] no [13:18:04] :D [13:18:24] (02:00:09 PM) YuviPanda: "JohannesK_WMDE: hmm, perhaps it can then run on its own labs instance? should be pretty easy to pick that up that way" [13:18:25] Coren: okiedokie :) [13:18:39] can't even find it :( [13:18:39] YuviPanda: catgraph *is* running on its own labs instance [13:18:45] what did you do to it... [13:18:50] Coren: thanks [13:18:55] hmm? why is it taking up memory on tools-labs? [13:19:24] petan: https://bugzilla.wikimedia.org/show_bug.cgi?id=49946 [13:19:52] Coren I did that... [13:20:03] Coren the bug contains 2 issues, you answered just to one of them [13:20:12] ... [13:20:13] ah one more question, whilst i'm here: on the old toolserver one had to specify a runtime-limit for a task, i didnt see anything on that for the new tool labs. so i just qsub all my tasks like so "qsub foo.sh" no matter the runtime? [13:20:18] oh. Didn't notice. Lemme go check. [13:20:29] nutz: Right. No such distinction here. [13:20:33] Coren: neat :) [13:20:35] thanks [13:20:35] petan: I go see [13:21:54] YuviPanda: what you want to do with graphs and neo4j may be already covered by catgraph. did you check it out? [13:22:03] ah, no. let me do that. [13:22:21] I've Caught Not Reading, will do :) [13:22:37] neo4j was just a random idea tho, currently doing something else [13:22:39] YuviPanda: catgraph isn't taking memory on tool labs. the article list generator backend is. :) [13:23:15] YuviPanda: http://tools.wmflabs.org/render-tests/catgraph/ (partly outdated... the instances on toolserver are still there, but the new stuff is on sylvester in labs) [13:23:48] heh, at this point I should just read up more [13:23:49] :) [13:25:03] * nutz waves at lbenedix [13:25:07] * nutz waves at lbenedix1 [13:28:47] valhallasw: addshore behold, stacktraces :) http://tools.wmflabs.org/dumpscan/cgi-bin/dumpscan.py [13:29:00] :D [13:29:18] petan: How odd. Something broke in puppet and the upstart_job resource doesn't seem to be working anymore; that left your boxen with a missing file. Patching around it. [13:29:29] ok [13:31:55] addshore [13:32:04] yus [13:32:06] bug #11 in pidgeon partially resolved [13:32:11] what dcc you need? :P [13:32:19] pidgeon now support ssl chat and regular chat in dcc [13:32:25] * addshore cant remember what 11 was ;p [13:32:47] pidgeonclient.org/bugzilla/show_bug.cgi?id=11 [13:32:51] http://pidgeonclient.org/bugzilla/show_bug.cgi?id=11 [13:33:08] :D [13:33:18] :o [13:33:20] i cant even remember when / why I needed it :> [13:33:27] but I didn't have a chance to test it :P [13:33:34] it's useful since NSA is scanning everything [13:33:51] SSL DCC is IMHO one of most secure means of communication on internet these days [13:33:52] is possible to read the my.cnf easily from php? [13:34:04] fale: no [13:34:14] petan: awesome :D thanks [13:34:17] my.cnf is not distributed by mysql server [13:34:46] fale, wait you mean your own .my.cnf? [13:34:47] petan: Fix't; but you'll need to run puppet and reboot [13:34:52] petan: I meaned the replica.my.cmf ;) [13:34:54] Coren ok [13:35:00] fale oh yes that is possible and very easy [13:35:09] fale: I thought you mean /etc/mysql/my.cnf [13:35:18] my.cnf != .my.cnf [13:35:23] petan: sorry [13:35:28] :) [13:36:25] addshore: http://tools.wmflabs.org/dumpscan/cgi-bin/dumpscan.py/index [13:36:49] > [13:36:50] :> [13:37:02] YuviPanda: if your good with python you could probably finish the scanning script also ;p [13:37:09] scroll to the bottom and youll see everything :. [13:37:24] then where would you get practice from? :P [13:37:35] I think I can consider this 'done', let me pull request [13:37:43] addshore: isn't this going to be super slow? [13:37:51] addshore: going through dumps instead of doing a mysql query? [13:37:58] addshore: does this tool already exist elsewhere? what is the purpose of this? [13:37:58] addshore come on, buglist is empty now :/ go fill in some bug don't tell me pidgeon is so perfect it doesn't have bugs :P [13:38:35] Coren I have to reboot all servers? [13:38:38] Coren: Poke? [13:41:21] Coren can I create a symlink in puppet git if I want 2 files to have same content? [13:41:29] I don't even know if git support symlinks [13:42:21] petan: git always follows symlinks... no way to turn that off afaik [13:43:51] YuviPanda: it would be interesting to combine that fulltext search with the deep, practically-instant category search of catgraph :) [13:43:54] JohannesK_WMDE: nonsense [13:43:59] git supports symlinks [13:44:04] valhallasw: how? [13:44:52] valhallasw: addshore https://github.com/addshore/dumpscan/pull/3 [13:45:01] let me know when you pull it, i'll set it up on tools [13:45:10] (requires virtualenv, which I've already setup) [13:45:12] http://bpaste.net/show/tWRFeeRl8ExkxdaxKZbW/ [13:45:17] JohannesK_WMDE, petan ^ [13:45:30] JohannesK_WMDE: true! I think some sortof pre-indexing for fulltext search is essential, though. Running through dumps would be madness [13:45:37] (I dunno what the current tool does, haven't looked at those yet0 [13:45:56] merged [13:46:01] JohannesK_WMDE: although I'm not sure what happens when you check it out on windows [13:46:07] !log toolsbeta petrb: rebooting all servers [13:46:09] addshore: that was fast :) [13:46:09] Logged the message, Master [13:46:19] Can I virtualenv from cgi? [13:46:35] definitely [13:46:43] virtualenv from everywhere \o/ [13:46:43] :) [13:46:46] a930913: just use #!/path/to/virtualenv/bin/python [13:46:49] valhallasw: yes, you can check in a symlink, but this is what happens to it: http://stackoverflow.com/questions/954560/what-does-git-do-to-files-that-are-a-symbolic-link (stores contents, not link) [13:47:03] addshore: look at https://github.com/addshore/dumpscan/blob/master/dumpscan.py for CGI + flask + virtualenv [13:47:06] valhallasw: Ta. [13:47:09] or just the first 3 lines for CGI + virtualenv [13:47:15] petan: Yes, reboot ALL THE SERVERS!!! [13:47:15] JohannesK_WMDE: http://bpaste.net/show/eHrJOArx54skn5xSmmGF/ [13:47:16] valhallasw: don't you need to add the site path too? [13:47:21] a930913: poke back! [13:47:29] YuviPanda: nope, that magically works as far as I know [13:47:33] I've created a tool that does need a mysql db (for his data). Does this db exists? I have to create it? On which hostname will responde? [13:47:38] hmm, it didn't for me when I was doing it for legoktm [13:47:49] JohannesK_WMDE: I have no clue what symlink support would mean other than 'when I clone the repo I get a symlink" [13:47:59] fale: You can create it, on 'tools-db'. [13:48:07] Coren: thanks [13:48:09] YuviPanda: valhallasw i still barely understand py;p [13:48:14] YuviPanda: hm, strange. Maybe there is something different from CGI, but I wouldn't know why [13:48:46] JohannesK_WMDE: also: we are in 2013 now, not 2009 ;-) [13:48:48] addshore: is this going to essentially 'grep' through the dumps? [13:48:59] JohannesK_WMDE: so it might just be that it has been added in the last four years [13:49:25] JohannesK_WMDE: eh, and your link actually states it just works?! [13:49:45] JohannesK_WMDE: "It then stores the name, mode and type (including the fact that it is a symlink) in the tree object that represents its containing directory." [13:49:55] valhallasw: the link says: "git just stores the contents of the link (i.e. the path of the file system object that it links to) in a 'blob' just like it would for a normal file. It then stores the name, mode and type (including the fact that it is a symlink) in the tree object that represents its containing directory." [13:50:12] valhallasw: which is exactly what i said before. [13:50:26] JohannesK_WMDE: yes. The contents of the *link*, not the contents of the *file* [13:50:45] that's not what 'following symlinks' means. [13:53:15] valhallasw: yeah, OK [13:53:40] addshore: also merge https://github.com/addshore/dumpscan/pull/4 [13:53:41] ? [13:54:42] petan: anyway, conclusion: you can make a symlink, and it will be cloned as a symlink [13:55:21] valhallasw, petan: it is ... different for symlinked directories though [13:56:17] valhallasw: That hasn't worked :( [13:56:34] JohannesK_WMDE: "It's worth noting that Pythonic's warnings about symlinked directories do not apply to versioned symlinks. The major edge case in question was that of folks symlinking some or all of the working tree into a different path (say onto a different partition with more disk space) and expecting git to check out code through the existing symlink." [13:56:56] valhallasw: do you have merge rights on addshore's dumpscan? [13:57:12] a930913: hm, sorry, I thought that should work. You can try the first three lines from https://github.com/addshore/dumpscan/blob/master/dumpscan.py , though [13:57:18] Coren: Might not need you if virtualenv can do what valhallasw said it can. [13:57:32] valhallasw: yes, i read that. see answer #54, same link. [13:57:52] "Any git pull with an update removes the link and makes it a normal directory" [13:58:18] YuviPanda: valhallasw closed the pull request just now [13:58:24] ty [13:58:34] sorry my internet just dropped :P [13:59:17] addshore, I'm teaching myself Objective-C [13:59:29] Hi all. Anyone know how to connect from tools-login to the tools web server? curl 'http://localhost/...' only says "couldn't connect to host". Tried "--noproxy localhost", or connection to the "full" URL, no joy. [13:59:49] magnus__: try tools-webserver-01 [14:00:25] magnus__: Johannes is hypothetically correct, but enduser access to the actual webservers isn't allowed for privacy policy reasons. [14:01:16]