[00:22:53] <Coren>	 labsdebrepo uses a directory name that's a potential clash with tool names, hence my use of .system
[00:23:49] <Coren>	 It's included in ::toollabs
[00:24:07] <Coren>	 file '/etc/apt/sources.list.d/local.list'
[00:24:36] <YuviPanda>	 Coren: yeah, figured it out. was just confused because the mongodb package from mongodb itself has a different name from the mongodb from ubuntu
[01:42:18] <wikibugs>	 3Wikimedia Labs / 3tools: Switch to using the mysqlnd driver in PHP - 10https://bugzilla.wikimedia.org/53117#c2 (10Krinkle) There's at least one I just ran into.  Functions like mysqli_result::fetch_all()[1] are only available with mysqlnd.  I'm not sure why that is, but reading different bits of documentati...
[01:43:03] <wikibugs>	 3Wikimedia Labs / 3tools: Switch to using the mysqlnd driver in PHP - 10https://bugzilla.wikimedia.org/53117#c3 (10Krinkle) Fatal error: Call to undefined method mysqli_result::fetch_all() in /data/project/orphantalk/src/mw-tool-orphantalk/lib/basetool/GlobalFunctions.php on line 181  Seems not an uncommon p...
[01:44:33] <wikibugs>	 3Wikimedia Labs / 3tools: Switch to using the mysqlnd driver in PHP - 10https://bugzilla.wikimedia.org/53117#c4 (10Yuvi Panda) I suppose the easiest way to do this is to just upgrade to PHP 5.4.
[01:51:48] <wikibugs>	 3Wikimedia Labs / 3tools: Document how to install Python modules in a tool's home directory/virtual environment - 10https://bugzilla.wikimedia.org/61824 (10Tim Landscheidt)
[01:54:35] <wikibugs>	 3Wikimedia Labs: WMFLabs: meta_p.wiki contains centralauth (unexpected wiki.url field is NULL) - 10https://bugzilla.wikimedia.org/65789 (10Krinkle) 3NEW p:3Unprio s:3normal a:3None I'm using code like the following as the basis for the <select> menu of wiki-agnostic tools.       $conn = kfGetMetaDb();...
[02:02:49] <wikibugs>	 3Wikimedia Labs / 3tools: Packages to be added to toollabs puppet - 10https://bugzilla.wikimedia.org/53704 (10Tim Landscheidt)
[02:02:49] <wikibugs>	 3Wikimedia Labs / 3tools: Install python module matplotlib - 10https://bugzilla.wikimedia.org/61445 (10Tim Landscheidt) a:5Marc A. Pelletier>3Tim Landscheidt
[02:02:50] <wikibugs>	 3Wikimedia Labs / 3tools: Install tmux on tools-login and tools-dev - 10https://bugzilla.wikimedia.org/65426 (10Tim Landscheidt) a:5Marc A. Pelletier>3Tim Landscheidt
[02:02:50] <wikibugs>	 3Wikimedia Labs / 3tools: install tesseract command line tool and the associated language package - 10https://bugzilla.wikimedia.org/65354 (10Tim Landscheidt) a:5Marc A. Pelletier>3Tim Landscheidt
[07:00:30] <GerardM->	 Coren YuviPanda|zzzzzz  andrewbogott_afk is there a problem
[07:00:34] <GerardM->	 Tools do not load
[07:03:42] <GerardM->	 things are back to normal it seems
[07:52:31] <petan>	 @token-off
[07:52:59] <labs-morebots>	 Logged the message, Master
[07:55:05] <petan>	 Damianz: why cluebot has 700+ GB of data in its folder?
[07:55:09] <petan>	 o.O
[08:44:04] <hashar>	 !log integration rebased puppetmaster
[08:44:06] <labs-morebots>	 Logged the message, Master
[09:56:42] <DanielK_WMDE__>	 hi all. for some reason, i can't start a web service for my tool. i'm getting this:
[09:56:49] <DanielK_WMDE__>	 /usr/local/bin/lighttpd-starter: line 45: cannot create temp file for here-document: No space left on device
[09:57:11] <DanielK_WMDE__>	 I do not see any full disks anywhere though
[09:57:20] <DanielK_WMDE__>	 what gives?
[09:59:28] <DanielK_WMDE__>	 bah.
[09:59:49] <DanielK_WMDE__>	 found http://bots.wmflabs.org/~wm-bot/logs/%23wikimedia-labs/20140516.txt  but doesn't seem to be related
[10:01:02] <DanielK_WMDE__>	 hm...
[10:01:23] <DanielK_WMDE__>	 when I try to start it again, it gives no error. but still does not run: http://tools.wmflabs.org/wikisense/
[10:01:27] <DanielK_WMDE__>	 wtf?
[10:04:10] <DanielK_WMDE__>	 now i get 2014-05-27 10:00:26: (configfile.c.1206) a default document-root has to be set
[10:04:13] <DanielK_WMDE__>	 this is confusing :)
[10:06:32] <DanielK_WMDE__>	 seems like the defaults don't get applied?...
[10:07:35] <DanielK_WMDE__>	 now we are back to line 45: cannot create temp file for here-document: No space left on device
[10:07:48] <DanielK_WMDE__>	 giving up for now. please ping me if you have any idea...
[10:09:20] <saper>	 nasty enviroment issues :/
[10:12:27] <scfc_de>	 / on tools-webgrid-01 is full, let's see why.
[10:19:17] <DanielK_WMDE__>	 scfc_de: is there no monitoring to tell folks when something like this happens?...
[10:19:29] <scfc_de>	 tools.zoomviewer is the culprit (4.6 GByte).  Let's see what looks no longer needed.
[10:19:59] <scfc_de>	 DanielK_WMDE__: No; Icinga (icinga.wmflabs.org) monitors the disk usage, but no alerts have been defined AFAIK.
[10:20:03] <DanielK_WMDE__>	 that thing presumably caches a lot of image data
[10:20:20] <DanielK_WMDE__>	 it's essentially a tile server, for hi res pictures, iirc
[10:20:36] <DanielK_WMDE__>	 scfc_de: sounds like that would be a good idea
[10:21:29] <DanielK_WMDE__>	 we should also have per-tool quotas, simply to keep shit like this from happening
[10:22:11] <DanielK_WMDE__>	 on the TS, everyone had a quota - if you need more space, just ask for it, but there was always a quota sto keep runaway scripts from filling up the disk and killing the ser4vice 
[10:22:21] <scfc_de>	 !log tools tools-webgrid-01: /tmp was full, removed files of tools.zoomviewer older than five days
[10:22:23] <labs-morebots>	 Logged the message, Master
[10:23:32] <scfc_de>	 DanielK_WMDE__: The problem here was that /tmp on the instance was filled.  If you specify a quota for that, it would be a few megabytes per tool at most.  dschwen should just put the temporary files beneath /data/project/zoomviewer where space is virtually limitless.
[10:24:08] <DanielK_WMDE__>	 one more reason to have a quota on /tmp
[10:28:23] <DanielK_WMDE__>	 scfc_de: webservice is up now, thanks
[10:29:24] <scfc_de>	 np
[10:34:33] <scfc_de>	 And full again?  *argl*
[10:36:02] <scfc_de>	 !log tools tools-webgrid-01: removed all files of tools.zoomviewer in /tmp
[10:36:04] <labs-morebots>	 Logged the message, Master
[10:44:58] <wikibugs>	 3Wikimedia Labs / 3tools: Allocate more space for /tmp on tools-webgrid-{01,02} - 10https://bugzilla.wikimedia.org/65801 (10Tim Landscheidt) 3NEW p:3Unprio s:3normal a:3Tim Landscheidt From time to time, /tmp on the webgrid nodes gets filled up by a tool or a core dump.  The instances have 150 GByte...
[14:40:07] <YuviPanda|zzzzzz>	 scfc_de`: saw the mongo patches? :)
[14:56:45] <bd808>	 !log deployment-prep Updated scap to 9609e8d
[14:56:48] <labs-morebots>	 Logged the message, Master
[15:03:26] <wikibugs>	 3Wikimedia Labs / 3deployment-prep (beta): [OPS] udp2log prevents udp2log-mw from starting - 10https://bugzilla.wikimedia.org/38995#c3 (10Antoine "hashar" Musso) 5RES/WOR>3REO It happens when the instance boot.  The udp2log service is started up on port 8420.  Then udp2log-mw attempt to start using the s...
[15:13:21] <JohannesK_WMDE>	 Coren: fyi, i ran into the same puppet problem i described yesterday. i found a solution here in the last paragraph: http://projects.puppetlabs.com/projects/1/wiki/certificates_and_security
[15:13:45] <JohannesK_WMDE>	 Coren: (basically, remove the whole ssldir on the client, then retry)
[15:13:58] <Coren>	 That seems...  violent.  :-)
[15:14:02] <JohannesK_WMDE>	 Coren: happened after changing the puppetmaster.
[15:14:21] <JohannesK_WMDE>	 well, that's what they suggested on the wiki. it worked. :)
[15:20:15] <phe>	 SGE question, I have trouble with qalter -notify, my job receive a SIGUSR2 but shutdown way to fast to finish to exec the sig handler, it's killed in a few milliseconds, the only things suspicious I a se is with qconf 
[15:20:28] <phe>	 qconf -sq continuous .... --> notify                00:00:60
[15:20:34] <phe>	 Is that a correct way to specify a 60 seconds delay between SIGUSR2 --> SIGKILL?
[15:30:04] <Coren>	 phe: I beleive it is; but that is somewhat contingent on /why/ the job is being shutdown.
[15:30:19] <Coren>	 phe: OOM has very little room.
[15:31:52] <phe>	 Coren, I want only it for qdel, qdel sigkill and no cleanup are annoying or some tool
[15:32:47] <phe>	 beside that, it's python, and signal in python is a mess...
[15:35:14] <Coren>	 qdel sigtems now; or at least it should.
[15:35:18] <Coren>	 sigterms*
[16:03:58] <renoirb>	 Hi guys, I was wondering if I'm missing something about Varnish and $u = User::newFromName('Myuser', 'creatable'); $u->load(); /* setters... */  $u->saveSettings(); $u>setCookies(); 
[16:08:24] <andrewbogott>	 Coren:  I have a new renaming problem today… changing a user's on-wiki name whilst leaving their shell name + associated bits intact.  Is that something mediawiki supports?
[16:08:24] <renoirb>	 Hi guys, I was wondering if I'm missing something about SSO use-case. Its about creating/loading and impersonating and the requirements when behind Varnish.   Roughly I have a OAuth Resource server that gives trusted data, I create a user like:  $u = User::newFromName('Myuser', 'creatable'); $u->load(); /* setters... */  $u->saveSettings(); $u>setCookies();   /* redirect */  and we are in business! But not behind Varnish..   Varnish file:
[16:08:24] <renoirb>	  https://github.com/webplatform/varnish-configs/blob/master/docs.vcl    Project:  https://github.com/webplatform/mediawiki-fxa-sso/tree/webplatform-sso
[16:09:07] * YuviPanda waves at andrewbogott
[16:09:15] * andrewbogott waves back
[16:09:18] <Coren>	 andrewbogott: That should work with the normal rename.
[16:09:46] <Coren>	 andrewbogott: Though it may leave the LDAP gcos field with the old value (mostly irrelevant, but might be worthwhile to fix anyways)
[16:09:58] <andrewbogott>	 ok, I think I don't know what 'the normal rename' is.
[16:11:20] <YuviPanda>	 andrewbogott: re: proxy, things seem to be stable since the 1.5 revert
[16:11:24] <YuviPanda>	 :( :)
[16:11:47] <YuviPanda>	 andrewbogott: I do have a full testing setup for the tools proxy setup at tools-proxy-test.wmflabs.org, so might try messing around with that at some point
[16:11:57] <YuviPanda>	 andrewbogott: might also consider upgrading that box to trusty and see if that has any effects
[16:12:05] <andrewbogott>	 YuviPanda: yeah, it would be great if you can reproduce the failures we saw with 1.6 in a testing env
[16:12:12] <andrewbogott>	 That gives us a chance of knowing if/when we can safely upgrade
[16:12:21] <YuviPanda>	 andrewbogott: yeah
[16:12:29] <andrewbogott>	 YuviPanda: you definitely can't upgrade the box in place, we'd have to build a new one.
[16:12:35] <YuviPanda>	 andrewbogott: yeah, that's what I meant. 
[16:12:44] <andrewbogott>	 'k
[16:12:54] <YuviPanda>	 andrewbogott: just remove that box, and just create a new one. it is trivial to setup a slave of redis, and it doesn't affect the master's perf much
[16:13:01] * andrewbogott nods
[16:13:29] <andrewbogott>	 Coren:  nm, found it
[16:15:48] <andrewbogott>	 Coren:  One other naive question… MW usernames are fully case sensitive?  So Andrew, anDrew and andreW could be three different people?
[16:16:12] <Cyberpower678>	 andrewbogott, except for the first letter.
[16:16:40] <Cyberpower678>	 andrewbogott, but yes, they are case sensitive.
[16:16:52] <Cyberpower678>	 cyberpower == Cyberpower
[16:17:04] <Cyberpower678>	 cyberpower != cyberPower
[16:17:15] <andrewbogott>	 Cyberpower678: ok, isn't https://wikitech.wikimedia.org/wiki/User_talk:Tim_Landscheidt#Account_creation_strangeness disagreeing with you?
[16:17:24] <andrewbogott>	 It sounds like he has two different accounts that differ only by first letter cap
[16:18:02] <legoktm>	 andrewbogott: MediaWiki will normalize them as the same, but the ldap extension that wikitech uses might do something funny
[16:18:25] <andrewbogott>	 :(
[16:18:47] <Cyberpower678>	 It appears to be an issue with the shell username.
[16:19:14] <andrewbogott>	 Cyberpower678: well, yes, the two different accounts have two different shell names, but...
[16:19:23] <andrewbogott>	 nevertheless, there are clearly two different mw accounts, one G and one g
[16:19:39] <Cyberpower678>	 Umm...
[16:19:44] <Cyberpower678>	 Hmm...
[16:20:18] <andrewbogott>	 If some bits of MW conflage G and g then when I rename I won't know which one I'm renaming
[16:20:23] <andrewbogott>	 *conflate
[16:20:56] * andrewbogott wonders if there's a special "1000 login accounts broken!" achievement barnstar
[16:22:19] <Cyberpower678>	 Hmm...Well going by the wikimedia engine, grafzahl will never be accessible.
[16:23:03] <legoktm>	 no it will, because authplugin (ldap) will take over first
[16:23:05] <andrewbogott>	 ok, so I probably need to start by deleting Grafzahl
[16:23:14] <andrewbogott>	 since grafzahl has the shell name he wants
[16:23:20] <Cyberpower678>	 When you type in a username, it sanitizes it first and then looks it up in the database.
[16:23:53] <andrewbogott>	 Although I don't know how I'll avoid deleting grafzahl by mistake
[16:24:00] * Cyberpower678 double checks something.
[16:24:18] <Cyberpower678>	 Damn session bug. 
[16:25:26] <Cyberpower678>	 I'm not sure how Semantic Wiki operates, but the normal wiki operates as I described above.
[16:26:05] <Cyberpower678>	 The safest way I see it is to take a special script and make the alterations directly to the database.
[16:26:48] <andrewbogott>	 In theory the username only appears in one place in the db, right?  In the users table?
[16:26:58] <Cyberpower678>	 Correct.
[16:27:08] <andrewbogott>	 and then of course I'll have to alter ldap as well to correspond.
[16:27:12] <Cyberpower678>	 rev_user_text will remane the same.
[16:27:25] <andrewbogott>	 ok, this seems possible… but I'll wait for the go-ahead from the user.
[16:27:28] <andrewbogott>	 thanks for explaining
[16:28:18] <Cyberpower678>	 AFAIK, when you look up cyberpower678 on wiki, it sanitizes the username and looks up Cyberpower678 on the database.
[16:28:47] <andrewbogott>	 Yeah, that mostly fits
[16:28:59] <andrewbogott>	 although it's still a mystery how things got into the state they're in
[16:29:31] <Cyberpower678>	 So the wikitech username grafzahl, not the shell, isn't accesible with the onwiki tools.
[16:30:08] <andrewbogott>	 yep, and that username is associated with the correct shell name.  So it's the one I should rename to g(G)rafZahl
[16:30:27] <andrewbogott>	 which should be fine, because at that point mw will think it's different from Grafzahl due to the Z
[16:30:36] <Cyberpower678>	 Yep.  It'll have to be done with a script AFAIK.
[16:31:51] <Cyberpower678>	 andrewbogott, make sure when renaming it in the DB it's a capital G.  Otherwise, wikitech won't recognize it.
[16:32:01] <andrewbogott>	 ok, cool
[16:33:48] <Cyberpower678>	 Look for any pages starting with grafzahl as they won't likely be accessible on Wikitech either, and rename them to.
[16:48:54] <Cyberpower678>	 Coren, there seems to be an issue with jsub
[16:49:13] <Cyberpower678>	 Coren, job deletion seems to have hung up
[16:49:30] <Cyberpower678>	 also, some of my tasks are sitting in a queue.
[16:49:48] <Cyberpower678>	 Which shouldn't be happening with my cyberbot node.
[16:52:36] <Cyberpower678>	 petan, ^^^
[16:54:56] <Cyberpower678>	 *sigh*
[16:55:03] <Cyberpower678>	 scfc_de`, ^^^
[17:12:19] <liangent>	 I'm curious why I got a bunch of Chinese spams sent to tools.toolname@tools.wmflabs.org
[17:42:37] <Ocaasi>	 Hi There.  I'm working on local-reference-api in Tool Labs.  It's pairing with this userscript: https://en.wikipedia.org/wiki/User:Nischayn22/OCLC.js We're trying to send an ip address to an API.
[17:42:59] <Ocaasi>	 The problem is, the $_SERVER['REMOTE_ADDR'] comes up as to 10.68.16.4 instead of the IP address of the client
[17:45:42] <Ocaasi>	 Could someone assist a bit?
[17:48:56] <Cyberpower678>	 Coren, ping again
[17:49:23] <rohit-dua>	 Ocassi: you cannot get the remote ip. its against the t&c
[17:51:27] <Ocaasi>	 rohit-dua: hmm.  that would be an issue.  this is an opt-in userscript with full disclosure about privacy and sharing permissions.  does that change anything if it only runs when people install it?
[17:52:49] <rohit-dua>	 Ocaasi: well i think we can get private info after users consent.  
[17:52:51] <phe>	 Coren, I think I found the trouble with SIGUSR2, qsub and jsub works fine, jsub -continuous kill too fast the process because with -continuous jsub create a wrapper script to run the job, the wrapper script doesn't trap signal, so SIGUSR2 is translated immediatly as SIGTERM (default action for non trapped SIGUSR2), the wrapper script die so on all its child before they have time to receive and handle SIGUSR2
[17:53:02] <Ocaasi>	 Or to ask it another way, do no other tools use local ip information to provide a service, even on an opt-in basis...?
[17:53:42] <Ocaasi>	 rohit-dua: yes, you'd have to actually install the userscript, and the documentation anywhere the install is suggested includes detailed privacy disclosures
[17:54:01] <Ocaasi>	 btw, WMF Legal has reviewed the disclosure
[17:54:07] <Ocaasi>	 and was ok with it
[17:54:55] <valhallasw>	 Ocaasi: I'm confused as to why you need the remote IP, though.
[17:55:04] <rohit-dua>	 Ocaasi: I had the same issue. most tools use cookies etc. but i  dont know how to get REMOTE_PORT  to get working
[17:55:39] <rohit-dua>	 REMOTE_IP  i mean
[17:55:43] <Ocaasi>	 valhallasw: maybe i'm not explaining well.  this tool sends the user's ip address to an external api.  that api checks to see if the ip address would get them access to full text of a reference in a university library.  if so, it returns a link to it.
[17:56:58] <Ocaasi>	 once it has the user's ip address, it sends that ip to a proxy resolver at the university library to check if it has access.  so access depends on having the user's ip address.
[17:57:27] <Ocaasi>	 i'm not sure if that's the 'remote ip' as i'm a bit of a technical noob
[17:58:11] <valhallasw>	 Yeah, that's what you would need. I'm trying to think of another way to get the IP, e.g. by letting the client browser add it to the request.
[17:59:11] <valhallasw>	 I'm not sure how to do that without remote_addr, to be honest :/
[18:00:09] <Ocaasi>	 i was looking at this page http://www.phpf1.com/tutorial/get-ip-address.html which explains how to get the ip address using php.  i'm not sure i understand why that doesn't work...
[18:00:48] <valhallasw>	 that still requires the server to have knowledge of the 'real' remote IP. The IP you're getting on the tools webserver is the internal proxy IP
[18:00:56] <Ocaasi>	 ah
[18:01:02] <valhallasw>	 which strips the real IP from the request
[18:01:30] <Ocaasi>	 so tools webserver strips the real ip?
[18:01:35] <valhallasw>	 yep
[18:03:04] <Ocaasi>	 is this tools webserver's way of protecting privacy and enforcing that aspect of the T&C, or would this be a problem with other webservers too?
[18:03:22] <YuviPanda>	 yup
[18:04:06] <Ocaasi>	 YuviPanda: sorry, yes to both of those questions?
[18:07:36] <YuviPanda>	 Ocaasi: just the first
[18:07:46] <Ocaasi>	 YuviPanda: thanks
[18:08:40] <Ocaasi>	 i realize i'm pushing this a bit, but with WMF approval, could there ever be a technical exception to that stripping?
[18:08:44] <Ocaasi>	 or is it hard-coded and impossible to change the behavior?
[18:08:49] <rohit-dua>	 Ocaasi: maybe you could use external stat services like <script type="text/javascript" src="http://l2.io/ip.js?var=myip"></script>
[18:08:59] <rohit-dua>	 http://l2.io/
[18:10:13] <Ocaasi>	 rohit-dua: thanks!  i will share that with the programmer
[18:10:21] <YuviPanda>	 rohit-dua: against the ToS that
[18:10:30] <YuviPanda>	 Ocaasi:   ^
[18:10:30] <Ocaasi>	 oh :/
[18:10:36] <Ocaasi>	 i see
[18:10:54] <YuviPanda>	 Ocaasi: you can get  a separate project and then get the ip that way
[18:11:17] <YuviPanda>	 Ocaasi: that is what UTRS does for example
[18:11:38] <Ocaasi>	 YuviPanda: interesting...what is a separate project?
[18:12:09] <YuviPanda>	 Ocaasi: you get your own machine and do what you like with it :)
[18:12:17] <Ocaasi>	 YuviPanda: ah! of course.
[18:12:52] <rohit-dua>	 YuviPanda: or he could find any free/open-souce pojects providing the same service.
[18:13:23] <YuviPanda>	 Ocaasi: rohit-dua i wasn't being clear, sorry. you could get a separate project *on* wikimedia labs.
[18:13:24] <Ocaasi>	 there are some other open access projects that may let us host this
[18:13:36] <YuviPanda>	 Ocaasi: rohit-dua it gives you your own machine you can do what you want with.
[18:13:45] <YuviPanda>	 Ocaasi: rohit-dua you can apply by poking andrewbogott :)
[18:13:57] <Ocaasi>	 YuviPanda: thanks much, super helpful!
[18:14:31] <Ocaasi>	 YuviPanda: is there any documentation on that process?
[18:14:40] <YuviPanda>	 Ocaasi: there is, but I can't find it at the moment.
[18:14:43] <YuviPanda>	 let me check again
[18:14:50] <Ocaasi>	 thank you kindly
[18:14:51] <petan>	 YuviPanda: what's the point of that IP restriction if anyone can just get own machine to get to it anyway
[18:15:09] <petan>	 this so much makes no sense
[18:15:25] <YuviPanda>	 petan: you don't get the nicenss of tools. if their machine is down they have to be the ones that figure it out and bring it back up
[18:15:57] <petan>	 eh, IMHO own machine is significantly more nicer than tools, but that still doesn't explain what is that restriction for, when you can easily bypass it?
[18:16:00] <YuviPanda>	 petan: plus this also means that projects can include scripts hosted on tools in them without violating privacy policy, while the same is not true for anything else hosted on labs.
[18:16:08] <YuviPanda>	 petan: well, that's just your opinion :P
[18:16:26] <YuviPanda>	 petan: that restriction is there so that tools' Privacy policy is compatible with wikimedia/wikipedia's?
[18:16:29] <petan>	 yes, maybe... but you know own machine = huge freedom, tool labs = lot of restrictions
[18:16:48] <YuviPanda>	 with freedom comes responsibility :)
[18:16:49] <petan>	 and why does it need to be compatible?
[18:17:01] <YuviPanda>	 <YuviPanda> petan: plus this also means that projects can include scripts hosted on tools in them without violating privacy policy, while the same is not true for anything else hosted on labs. 
[18:17:36] <petan>	 but "anything else" hosted on labs is being used in same fashion as tool labs tools
[18:17:55] <petan>	 you can just take your tool from tool labs, get own instance and host it there, link it from wikipedia just the same
[18:18:17] <petan>	 there are 3rd tools linked on wikipedia that aren't even hosted on wmf servers
[18:18:27] <YuviPanda>	 like what, exactly?
[18:18:31] <petan>	 let me find it
[18:18:52] <petan>	 open https://en.wikipedia.org/w/index.php?title=Main_Page&action=history
[18:19:08] <YuviPanda>	 yeah?
[18:19:11] <petan>	 now you see a link · Revision history search hosted on ramselehof.de
[18:19:23] <YuviPanda>	 no?
[18:19:25] <petan>	 ramselehof.de isn't AFAIK hosted by wmf :P
[18:19:26] <YuviPanda>	 I don't
[18:19:39] <petan>	 it's on line "external tools"
[18:19:44] <petan>	 you can also hit ctrl + f
[18:19:50] <YuviPanda>	 petan: right, they are being *linked to*
[18:19:58] <YuviPanda>	 petan: very different from being included in
[18:20:02] <petan>	 yes, that is how tool labs tool are used
[18:20:16] <petan>	 they are also just linked to, I don't know single tool labs tool which is literally "included"
[18:20:17] <YuviPanda>	 either way, what are you suggesting? :)
[18:20:31] <Cyberpower678>	 petan, hi.  There's an issue with jsub, at least on the cyberbot exec node.
[18:20:36] <petan>	 I am suggesting "get rid of that ridiculous restrictions, freedom ftw"
[18:20:46] <petan>	 YuviPanda ^
[18:20:52] <petan>	 dictatorship suck
[18:20:56] <petan>	 Cyberpower678: hi
[18:21:10] <petan>	 or * dictature, however you say that in english
[18:21:30] <Cyberpower678>	 petan, for some reason, the scheduler hung up.  I can't kill any jobs and jobs I schedule just sit there.
[18:21:37] <petan>	 hmm
[18:21:41] <petan>	 let me check
[18:21:44] <YuviPanda>	 petan: I personally have no opinions on ip visibility, but good luck with getting rid of that :)
[18:22:11] <petan>	 that restriction makes no sense... nobody was able to explain to me a benefit of having it
[18:22:49] <YuviPanda>	 we also couldn't explain to you why rewriting memcached in C# was a bad idea, or the security holes in one of the other tools you wrote.
[18:22:59] <rohit-dua>	 the tools table in  http://tools.wmflabs.org/  shows links to tools homepage right? But it does not do it for mine.
[18:23:11] <rohit-dua>	 also  my [HOMEPAGE] index.py is inside public_html/app/index.py
[18:23:37] <Cyberpower678>	 rohit-dua, try moving it to the parent folder.
[18:24:10] <petan>	 YuviPanda: not true, after some time I finaly found that security issue, regarding memcached, it took a while for guys on Stack Overflow to explain to me that C# is too high level language to implement anything like memcached in, I still don't think it was a bad idea to rewrite it, just the language choice was wrong :P
[18:24:16] <rohit-dua>	 Cyberpower678: cant it get working from the subfolder by changing some settings/ or something
[18:25:11] <petan>	 I was still hoping that c# give you some access to memory control but meh... it doesn't allow you to do much just as any other high level language
[18:25:12] <Cyberpower678>	 IDK.  I've got minimal experience when it comes to that.  A workaround is a redirect index page in public_html that redirects to app
[18:25:39] <YuviPanda>	 Ocaasi: can't find docs atm :( poke andrewbogott or Coren?
[18:25:43] <YuviPanda>	 Ocaasi: sorry! 
[18:25:54] <petan>	 Cyberpower678: can you elaborate a bit
[18:26:00] <Ocaasi>	 YuviPanda: no problem, thanks for all your assistance :)
[18:26:09] <petan>	 Cyberpower678: what actually doesn't work? I see a number of jobs running on your serve
[18:26:22] <petan>	 ah you can't kill them
[18:26:26] <YuviPanda>	 Ocaasi: yw :) feel free to email me as well if nobody's on irc 
[18:26:27] <petan>	 interesting
[18:26:35] <Ocaasi>	 YuviPanda: cheers
[18:26:43] <petan>	 Cyberpower678: give me a number a task which can be killed
[18:26:45] <YuviPanda>	 Ocaasi: yw :)
[18:26:48] <Cyberpower678>	 petan, well when I submit a job using jsub, it just sits in the queue waiting to start.  When I try to delete a job, it continues running waiting for deletion.
[18:27:00] <petan>	 ok
[18:27:16] <valhallasw>	 petan: the benefit is that it makes it significantly harder to get user IPs
[18:27:24] <Cyberpower678>	 I'm not sure if there's a bug with the cyberbot exec node or something else.
[18:27:31] <petan>	 ssh tools-exec-cyberbot.eqiad.wmflabs
[18:27:33] <petan>	 ssh: Could not resolve hostname tools-exec-cyberbot.eqiad.wmflabs: Name or service not known
[18:27:34] <valhallasw>	 which means it gets harder to violate the privacy rules
[18:27:37] <petan>	 that would be it ^
[18:27:51] <Cyberpower678>	 Umm what?
[18:27:58] <Cyberpower678>	 It was there when we migrated
[18:28:13] <Cyberpower678>	 Coren, ^^^
[18:28:29] <petan>	 valhallasw: really? define "significantly" it seems to me pretty easy to just ask for own box to run my service on, where it works the same, just it can read the IP just fine
[18:28:37] <valhallasw>	 rohit-dua: as far as I know, just having the public_html dir should be enough to get a link on the front page, but you could try adding an index.php that just redirects to the python index
[18:28:46] <valhallasw>	 petan: yes, it's easy enough for *you*
[18:28:56] <valhallasw>	 however, you are not the typical tools labs user
[18:29:12] <petan>	 I think there are many people out there would understand linux much more than me, even on labs
[18:29:17] <petan>	 * who
[18:29:47] <petan>	 so, I think it's pretty easy to bypass this restriction to many people, which makes it ineffective
[18:29:59] <Cyberpower678>	 petan, then how are my other tasks currently running.  They're all running on cyberbot-exec
[18:30:14] <valhallasw>	 no, it's actually not easy to bypass it for most people. Most people even have trouble just logging in to tools labs.
[18:30:18] <Cyberpower678>	 tools-cyberbot-exec
[18:30:28] <Cyberpower678>	 tools-exec-cyberbot
[18:30:42] <valhallasw>	 a block does not have to block everything to be effective
[18:30:42] <petan>	 ok I see these people would be very dangerous if they had access to my IP :P
[18:30:42] <valhallasw>	 case in point: DRM
[18:31:05] <valhallasw>	 it's not about security, it's about privacy.
[18:31:06] <petan>	 Cyberpower678: can you give me a NUMBER
[18:31:35] <petan>	 now you talk like EU :P
[18:31:45] <petan>	 IP isn't your credit card number
[18:31:47] <Cyberpower678>	 324236 can be killed without causing issues
[18:32:57] <petan>	 Cyberpower678: you are correct
[18:32:58] <valhallasw>	 petan: again, your comparison to a credit card number is completely bizarre. It's about *privacy*, not about *security*.
[18:33:03] <petan>	 it really is broken a lot :P
[18:33:19] <valhallasw>	 privacy is not about people stealing money from me, it's about people logging data about me
[18:33:38] <valhallasw>	 it's about someones right not to be tracked without consent
[18:33:42] <petan>	 I would say something more bizzare now, but I would have to kickban myself from channel after doing that
[18:33:52] <valhallasw>	 and it's fine if *you* don't care about *your* privacy
[18:34:19] <valhallasw>	 but I do care about mine, and more importantly, about other people's privacy.
[18:34:31] <petan>	 in fact I do care about privacy of everyone that's why I highlight that people should rely on this "privacy policy" because it's fragile and can be easily bypassed
[18:34:38] <valhallasw>	 and even though the IP 'blockade' is imperfect, it does improve privacy.
[18:35:01] <petan>	 it would be more secure if it didn't exist, then people would be more careful because they wouldn't live in a lie thinking their privacy is secured
[18:36:02] <valhallasw>	 I would be fine with not telling users we don't store IPs.
[18:36:10] <Cyberpower678>	 petan, so what's causing it.
[18:36:45] <petan>	 that I am investigating now
[18:37:00] <Cyberpower678>	 It apparently still worked 24 hours ago.
[18:37:50] <petan>	 valhallasw: but that's a lie. truth is: "we tell people that they should not store your IP but they may not be that kind to follow this requirement, so maybe some do store your IP".
[18:38:41] <valhallasw>	 Truth is: "On tools labs, we have implemented measures to stop people from storing your IP. However, this may not be the case on other Labs projects."
[18:39:21] <petan>	 it may not be the case of tool labs either, what if someone is breaking TOS by including 3rd server script to collect your IP?
[18:39:40] <petan>	 until you figure out they break TOS they collect hundreds of IP's
[18:40:02] <valhallasw>	 Yep. That's what happens with rules.
[18:40:07] <valhallasw>	 Welcome to the imperfect real world.
[18:40:22] <rohit-dua>	 how can i add a link(in bw text) in my tools description.?
[18:40:30] <petan>	 if people care about privacy they should buy a vpn, it's not so expensive
[18:40:56] <petan>	 some anonymous recommended me some sweden vpn and it's really cool :P
[18:40:58] <valhallasw>	 Spoken like a true first world citizen.
[18:41:32] <legoktm>	 speaking of privacy, today's http://www.smbc-comics.com/ is relevant
[18:41:38] <petan>	 NSA watches you anyway, these privacy policies won't help you
[18:42:10] <petan>	 legoktm: haha
[18:42:19] <petan>	 that's true
[18:42:36] <petan>	 by limiting access to IP's you are making some tools less awesome
[18:42:45] <valhallasw>	 Yes, and I've got nothing to hide. Anything else you'd like to add?
[18:43:04] <petan>	 for example awstats suck without ip's
[18:43:30] <valhallasw>	 and awstats serves which goal exactly?
[18:43:35] <petan>	 valhallasw: if you got nothing to hide why you care about privacy? :P
[18:43:52] <valhallasw>	 petan: Your sarcasm detector might be broken.
[18:43:59] <petan>	 valhallasw: they provide some nice statistics, like which countries visit labs most, in which time, etc
[18:44:26] <petan>	 I am completely broken
[18:45:43] <petan>	 Cyberpower678: problem is that your instance can't be resolved
[18:45:49] <petan>	 its record might been removed from dns server?
[18:46:19] <Cyberpower678>	 It must have happened sometime in the last 18 hours.
[18:46:22] <Cyberpower678>	 petan, ^
[18:46:28] <petan>	 hmm
[18:46:33] <mwalker>	 bd808|LUNCH (or gwicke), when you get back; I'm wondering if you've ever seen a status return like "fetch status: 0 [started: 1341 mins ago, last-return: 1341 mins ago]" from git deploy? This is for a fetch job that I just tried (but apparently there's one already running!?)
[18:46:54] <mwalker>	 *and this is here because this is on deployment-pdf01 host 
[18:46:58] <petan>	 Cyberpower678: I can fix it but not permanently
[18:47:03] <petan>	 I have no access to dns server
[18:47:24] <gwicke>	 mwalker, I don't remember seeing that particular error before
[18:48:11] <petan>	 scfc_de`, you know if /etc/hosts are in puppet?
[18:48:59] <Cyberpower678>	 petan, so what are you doing?
[18:49:29] <labs-morebots>	 Logged the message, Master
[18:55:53] <petan>	 Cyberpower678: fixed!
[18:57:14] <petan>	 tbh I don't even know if it was me who fixed it
[18:57:20] <petan>	 because it seems dns works now too
[18:58:14] <Cyberpower678>	 petan, well regardless, cyberbot exec is working again.
[18:58:30] <Coren>	 petan: /etc/hosts is generated by maintain-replicas
[18:58:38] <Coren>	 I.e.: outside labs.
[18:58:41] <bd808>	 mwalker: `fetch status: 0` should be the success status.
[18:58:53] <petan>	 Coren: ok, then I don't need to care about removing that line from it :P
[18:58:57] <Cyberpower678>	 He's awake. :-)
[18:59:15] <bd808>	 `git deploy report --detailed sync` should tell you the current state of the cluster
[18:59:18] <Coren>	 I'm awake.  Was at lunch followed by bank run.
[18:59:27] <petan>	 you robbed a bank o.O
[18:59:42] <mwalker>	 bd808, aye; it looks like salt didn't clean it's local /var/cache/salt/minion/proc directory
[18:59:48] <mwalker>	 so I just rm'd the heck out of it
[19:00:00] <mwalker>	 because the master didn
[19:00:06] <mwalker>	 t have any record of still running jobs
[19:00:36] <bd808>	 mwalker: Yuck. I still don't completely understand how salt does what it does.
[19:00:45] <mwalker>	 apparently neither do I
[19:00:50] <mwalker>	 because it still thinks it started a job
[19:00:54] <mwalker>	 days ago
[19:00:56] <mwalker>	 that didn't finish
[19:01:57] <bd808>	 The report command reads data from the redis instance on deployment-salt. It's persistent between runs.
[19:02:09] <mwalker>	 damnit
[19:02:17] <bd808>	 Is `git deploy start` not working?
[19:02:19] <mwalker>	 no
[19:02:46] <mwalker>	 `git deploy start`, `git deploy sync --force` is failing because it has this old job
[19:03:32] <bd808>	 mwalker: Have you tried `git deploy abort`?
[19:03:39] <mwalker>	 yep
[19:03:48] <mwalker>	 and git deploy finish
[19:04:53] * bd808 scans [[wikitech:Trebuchet]] for other ideas
[19:05:26] <bd808>	 You could try calling the runner directly from deployment-salt.
[19:05:47] <ori>	 You are in an open field west of a big white house with a boarded front door. There is a small mailbox here.
[19:05:49] <ori>	 > git deploy start
[19:05:57] <ori>	 I don't understand that.
[19:06:04] <bd808>	 `salt-run deploy.fetch 'ocg/ocg'`
[19:06:18] <bd808>	 `salt-run deploy.checkout 'ocg/ocg'`
[19:06:21] <ori>	 > look at git
[19:06:32] <bd808>	 ori: You are eaten by a grue
[19:06:45] <bd808>	 ori: Maybe light a candle next time
[19:07:02] <ori>	 :)
[19:07:07] <bd808>	 xyzzy
[19:07:07] <Cyberpower678>	 Coren, has Ganglia been moved to eqiad yet?
[19:08:04] <bd808>	 !log deployment-prep Updated scap to 48c7e28
[19:08:05] <labs-morebots>	 Logged the message, Master
[19:10:01] <mwalker>	 bd808, well; that salt-run finished correctly; but I'm going to go off in search of food before continuing this adventure
[19:10:09] <Coren>	 Cyberpower678: I believe it has.
[19:10:49] <Coren>	 bd808: A hollow voice intones "Fool!"
[19:16:47] <petan>	 Cyberpower678: which one
[19:17:08] <Cyberpower678>	 petan, ?
[19:17:15] <petan>	 production one?
[19:17:22] <petan>	 ganglia server
[19:17:34] <petan>	 everything from labs was already move
[19:17:35] <petan>	 d
[19:46:26] <wikibugs>	 3Wikimedia Labs / 3Infrastructure: Inconsistent GID number (500 or 550) - 10https://bugzilla.wikimedia.org/63028#c3 (10Andrew Bogott) 5NEW>3RES/FIX OK, done:  1)  Everyone with a primary group of 550 now has a primary group of 500 instead 2)  Everyone who used to have a primary group of 550 is now an exp...
[20:31:22] <awjr>	 who should i talk to about getting someone added to the tools project?
[20:32:33] <andrewbogott>	 awjr: I can add you.  Hang on...
[20:32:59] <awjr>	 andrewbogott: thanks - im already a maintainer of a couple tools on tools labs, but i want to add jon robson to be a maintainer of one of my tools
[20:33:16] <andrewbogott>	 ah, ok -- do you know what his wikitech username is?
[20:33:24] <andrewbogott>	 Oh, he's on the list already.
[20:33:41] <andrewbogott>	 Or at least 'Jdlrobson' is
[20:33:42] <awjr>	 orly? he wasn't showing up when i tried
[20:33:46] <awjr>	 lemme try again
[20:34:32] <awjr>	 andrewbogott: im an idiot, i was trying to add him via 'service user' rather than the 'member' field :)
[20:34:37] <awjr>	 thank you :)
[20:34:44] <andrewbogott>	 sure!
[20:35:28] <wikibugs>	 3Tool Labs tools / 3tsreports: Clean up old toolserver.org references - 10https://bugzilla.wikimedia.org/65832 (10Merlijn van Deen) 3NEW p:3Unprio s:3normal a:3Merlijn van Deen https://github.com/valhallasw/tsreports/search?q=toolserver&ref=cmdform
[20:36:04] <Coren>	 andrewbogott: I got the second copper pair on labstore; I'll be doing the bonding on Friday as planned.  No objections?
[20:36:12] <scfc_de>	 liangent: We had some bounces of possible spam recently (I don't speak Chinese :-)).  They seemed to come from non-existing domains.  Do you have example sender addresses of confirmed spam you received?
[20:36:18] <Coren>	 Moar bandwidth!!1!!one!
[20:36:21] <andrewbogott>	 Coren:  nope, sounds good.
[20:39:24] <hedonil>	 Coren: 10 GbE link adapters for OpenStack hosts? or more datacenter power?
[20:39:56] <andrewbogott>	 hedonil: bandwidth for the connection between VMs and nfs storage
[20:40:21] <hedonil>	 andrewbogott: yeah
[20:42:11] <hedonil>	 cubic capacity can only be beaten by even more cubic capacity, no scam with rotation speed :P
[20:43:34] <Coren>	 andrewbogott: paravoid does have legit concerns that usage will simply grow to match; but I can't think of a way to limit per-user.
[20:43:58] <Coren>	 andrewbogott: In fact, I don't know of a filesystem that allows that period.
[20:55:06] <andrewbogott>	 seems like if usage grows to match I might consider that a success rather than a failure
[20:55:13] <andrewbogott>	 unless we have reason to think it's frivolous usage
[21:20:11] <wikibugs>	 3Wikimedia Labs / 3deployment-prep (beta): beta labs mysteriously goes read-only overnight - 10https://bugzilla.wikimedia.org/65486#c5 (10James Forrester) FWIW VisualEditor doesn't know about <readonlytext> – it's just passing on what it gets from the API.
[21:26:01] <JohannesK_WMDE>	 Coren: still there?
[21:27:56] <wikibugs>	 3Wikimedia Labs / 3deployment-prep (beta): beta labs mysteriously goes read-only overnight - 10https://bugzilla.wikimedia.org/65486#c6 (10Chris McMahon)  Right, and we know about the unexpected readonly status because it seems only VisualEditor displays that error in a javascript confirm modal dialog.  It mi...
[22:14:41] <wikibugs>	 3Wikimedia Labs / 3deployment-prep (beta): Users with primary group of 550(svn) cannot sudo as mwdeploy on deployment-bastion - 10https://bugzilla.wikimedia.org/65548 (10Bryan Davis) 5PAT>3RES/FIX
[22:58:56] <wikibugs>	 3Wikimedia Labs / 3deployment-prep (beta): beta cluster job runner keep running some periodic tasks - 10https://bugzilla.wikimedia.org/63681#c6 (10Greg Grossmeier) p:5Unprio>3Normal (still seeing those log entries)
[23:56:38] <legoktm>	 I just booted up an instance that hasn't been touched since pre-eqiad migration, and noticed that on the Image ID it says "ubuntu-12.04-precise (deprecated 12-16-2013)"
[23:56:44] <legoktm>	 should I do anything about that?
[23:56:52] <legoktm>	 (the instance is social-tools1)
[23:58:33] <legoktm>	 andrewbogott_afk: ^
[23:59:22] <scfc_de>	 legoktm: I think all image types that aren't the "last" ones get that tag, so "deprecated" doesn't necessarily mean "full of errors".