[01:47:21] [bz] (8NEW - created by: 2m.p.roppelt, priority: 4Unprioritized - 6normal) [Bug 51212] libgcc_s.so.1 must be installed for pthread_cancel to work - https://bugzilla.wikimedia.org/show_bug.cgi?id=51212 [02:15:28] i was fired recently (as in terminated from a job). why did the boss ask me if i expected it and why does he and my colleagues still have me on skype? [02:15:59] Ummm. [02:16:20] have you ever been fired t13? [02:16:27] I have. [02:17:14] how did you act at the getting fired meeting/? [02:18:21] I walked out, talked to my friends and family, then found another job. Life goes on. [02:19:13] did they ask you about getting fired at the next interview? did you see it coming? [02:19:38] Nope. Yep. [05:19:10] [bz] (8REOPENED - created by: 2Chris McMahon, priority: 4Unprioritized - 6major) [Bug 50623] Entering AFTv5 feedback causes error - https://bugzilla.wikimedia.org/show_bug.cgi?id=50623 [05:19:25] [bz] (8NEW - created by: 2Chris McMahon, priority: 4Unprioritized - 6major) [Bug 50622] Special:NewPagesFeed error - https://bugzilla.wikimedia.org/show_bug.cgi?id=50622 [07:43:20] [bz] (8VERIFIED - created by: 2Antoine "hashar" Musso, priority: 4Unprioritized - 6normal) [Bug 51068] ganglia.wmflabs.org is missing most projects - https://bugzilla.wikimedia.org/show_bug.cgi?id=51068 [08:32:05] can a db on tools-db be shared between users ? :> [08:32:19] / between tools [09:35:07] addshore: yes [09:35:19] addshore: you need to name it blafvdfs_p [09:35:24] the _p [09:35:25] ahh, okay :> [09:40:47] petan: Why's blafvdfs? [09:41:01] example :P [10:25:14] petan: Finally got my cloak [10:25:56] Where's Cyberpower678? [11:53:12] @seen Cyberpower678 [11:53:12] zhuyifei1999: Last time I saw Cyberpower678 they were quitting the network with reason: Quit: Leaving N/A at 7/12/2013 7:21:39 AM (04:31:32.7196070 ago) [11:53:27] @notify Cyberpower678 [11:53:27] I'll let you know when I see Cyberpower678 around here [12:13:27] Cyberpower678: I got a cloak! [12:13:36] nice [12:13:52] !accesslist [12:13:52] http://helpmebot.org.uk/wiki/Special:AccessList [12:14:15] !access del 254 [12:14:16] Successfully removed access [12:14:16] Access entry removed. [12:15:14] !myaccess zhuyifei1999!~pidgeon@wikimedia/zhuyifei1999 [12:15:15] zhuyifei1999!~pidgeon@wikimedia/zhuyifei1999 has the access level: Advanced [12:15:27] Very good. [12:16:07] Cyberpower678: thanks [12:18:05] Who was 254? [12:18:23] me [12:18:30] Ahh. [12:19:18] If I had known that was what you wanted cp for, I would've done it an hour ago.. :p [12:20:09] Technical_13, I didn't know you were in here. [12:20:33] !myaccess Technical_13!~T13@wikipedia/Technical-13 [12:20:33] Technical_13!~T13@wikipedia/Technical-13 has the access level: Superuser [12:20:52] zhuyifei1999, ^ he's got the same rights as I do. [12:21:02] * Technical_13 rarely actually sleeps [12:21:22] * Cyberpower678 slept for ten hours last night. [12:21:46] Technical_13, 254 zhuyifei1999 % % Advanced 2013-07-03 13:16:31 [12:21:59] * zhuyifei1999 doesn't sleep that long [12:22:22] That was the access entry. He didn't have a cloak yet, so I gave his nick advanced rights. [12:29:20] Cyberpower678: Why's there unaffiliated/petan? [12:29:47] where [12:30:17] 125 [12:33:36] petan, can I deleted that or would you like to keep that? [12:35:42] I was going to ask petan why he isn't superuser myself but kept forgetting.. [12:39:08] what what [12:39:58] petan: why do you only have advanced Helpmebot privileges instead of superuser? [12:40:32] Curious as to you being manually added with an unaffiliated cloak. [12:42:45] Technical_13: why are there two bots again? [12:43:09] Only two? lol [12:45:14] YuviPanda use @kick? :P [12:45:22] @kick Helpmebot [12:45:33] poor bot :( [12:46:18] petan: why kick it? [12:46:27] Guess Helpmebot didn't like that... [12:46:28] zhuyifei1999 I didn't do that :D [12:46:35] YuviPanda did [12:46:46] zhuyifei1999: why have two bots? [12:46:55] does Helpmebot do something that wm-bot does not? [12:46:58] you told him (12:45:16) YuviPanda use @kick? :P [12:47:02] !lick YuviPanda [12:47:02] Helpmebot, comes back automatically. [12:47:03] * Helpmebot licks YuviPanda [12:47:08] it can do this ^ [12:47:14] meh [12:47:18] !smoke YuviPanda [12:47:19] !charge YuviPanda [12:47:20] * Helpmebot is chargin mah lazer, aims and fires at YuviPanda. [12:47:23] that makes it helpful :D [12:47:27] !smoke YuviPanda [12:47:43] mah lazer? [12:47:50] :D [12:47:55] Cyberpower678: !smoke YuviPanda not worjing? [12:48:03] *working [12:48:28] Because it doesn't know the command. [12:48:32] YuviPanda: you going to kick ChanServ, wm-bot, gerrit-wm, labs-morebots, jeremyb, and retsreklawts too? They are all bots... [12:48:52] Technical_13: ChanServ, wm-bot, gerrit-wm, labs-morebots all have different functionalities [12:49:04] Technical_13: and if they overlap I think they should be kicked [12:49:36] Cyberpower678: it was working [12:49:48] What was smoke? [12:49:57] !learn @action smoke marinates {0}, smokes them, and eats them afterwards. Munch Munch. [12:50:01] wait jeremyb is a bot? :D [12:50:06] finding log.... [12:50:16] !smoke zhuyifei1999 [12:50:17] Yes petan. [12:50:17] * Helpmebot marinates zhuyifei1999, smokes them, and eats them afterwards. Munch Munch. [12:51:05] ChanServ at some point overlaps with wm-bot [12:51:08] let me kick it [12:51:25] kick ChanServ :D [12:51:26] [5:12:35] !smoke wm-bot [12:51:27] [5:12:35] * Helpmebot sets wm-bot over hickory smoke and waits for the taste to be just right, then eats wm-bot. [12:51:33] done [12:51:40] @kick petan overlaps with wm-bot :) [12:51:50] @kick yuvipanda pre-emptive kick [12:52:17] So much abuse of admin powers. :p [12:52:20] hmm, friday evenings are a silly time [12:52:30] evening for you [12:52:33] here it's silly all day [12:52:38] !pat petan [12:52:38] * Helpmebot pats petan like a dog. *W00F* [12:52:38] https://labsconsole.wikimedia.org/wiki/Git#Fixing_a_path_conflict [12:52:38] true true [12:52:50] ^ overlap [12:52:53] and annoying, too [12:52:59] Cyberpower678: [12:53:00] [5:12:35] !smoke wm-bot [12:53:01] [5:12:35] * Helpmebot sets wm-bot over hickory smoke and waits for the taste to be just right, then eats wm-bot. [12:53:02] (Time in UTC, logged by YiFeiBot on 30130705) [12:53:15] !smoke zhuyifei1999 [12:53:15] * Helpmebot marinates zhuyifei1999, smokes them, and eats them afterwards. Munch Munch. [12:53:34] we should make #wikimedia-labs-offtopic [12:54:31] Yes, there is overlap YuviPanda. You can always !curl Helpmebot is it bothers you.. [12:54:51] Hmb has other functions that are useful. [12:54:57] like? [12:54:57] !helpmebot [12:54:58] * Helpmebot v6 is maintained by stwalkerster | My source code is available at http://is.gd/E7XrEy | The list of hard-coded commands can be found at http://is.gd/GQLN1a | The list of commands I have learned in-channel can be found at http://is.gd/hueY0E and may change at any time | The list of users with special access can be found at http://is.gd/AR00iW | I'm glad to be here to help you! [12:54:58] liar [12:55:15] ^^^ read up. :) [12:55:30] !wm-bot [12:55:30] http://meta.wikimedia.org/wiki/WM-Bot [12:55:30] liar [12:55:37] :D [12:55:47] !join #wikimedia-labs-offtopic [12:56:55] @trusted [12:56:55] 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), [13:00:00] Steinsplitter fixed :P [13:00:16] :) [13:02:24] YuviPanda can be happy now :D [13:02:30] no more spam in here [13:02:45] @infobot-link+ #wikimedia-labs-offtopic [13:02:56] !curl [13:02:56] * Helpmebot curls up in a ball. [13:03:04] @infobot-share-trust+ #wikimedia-labs-offtopic [13:03:04] You inserted channel #wikimedia-labs-offtopic to shared db [13:04:39] !help [13:04:39] !documentation for labs !wm-bot for bot [13:04:39] If you are trying to get help for a question you have related to the wiki, please type: !helper and your question. If you are trying to get help for one of my commands, you can find full details using !commands and following the link. [13:04:42] .. [13:04:47] !q Helpmebot [13:04:47] There are multiple keys, refine your input: queue, quilt, [13:04:51] @q Helpmebot [13:04:55] !help [13:04:55] !documentation for labs !wm-bot for bot [13:04:59] here we go [13:05:28] !forget help [13:05:51] !helpmebot [13:05:51] liar [13:06:56] @trusted [13:06:56] 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), [13:07:16] @unq Helpmebot [13:07:16] Permission denied [13:07:20] ... [13:07:32] '@' trusted does not list trusted channels, only users. Every time you do that everybody gets pinged. [13:07:34] do we really need Helpmebot here? :P [13:07:56] I guess not.. [13:08:09] Unq and I'll make it leave. [13:08:20] @unq Helpmebot [13:08:28] !quit [13:08:34] !part [13:08:46] o/ [13:11:48] YuviPanda did you have a look on that sul thing [13:12:00] petan: not yet petan. dunno when I can get to that [13:12:03] ok [13:12:15] addshore: how's your db migration? [13:12:24] mhhm [13:12:29] got some problems with my encoding again [13:13:05] petan: any progress on "all" not returning the proper $url? [13:13:28] Or new pages not showing up? [13:13:54] basically petan i had a mess of strings encoded, \u01e3\xf0 [13:14:01] i have the \u decoding nailed [13:14:09] just cant make php work with the \x bit :> [13:14:26] ok [13:14:29] im remembering to format my new table in utf8 of course ;p [13:14:30] Technical_13 nope [13:15:03] :( [13:19:06] !trout Technical_13 [13:19:18] !uncurl [13:19:19] Done [13:19:32] can we just keep it in -offtopic? [13:19:50] this channel should really be used only for labs related discussion [13:19:51] petan, We can use it to trout Coren if breaks labs. [13:20:16] - Oppose [13:20:35] I afraid that only person who frequently breaks labs is me ;) [13:20:38] Ēsatrēoƿ\xfe [13:20:41] Ēsatrēoƿ\xfe [13:20:51] addshore: indeed [13:21:16] how can i convert the \xfe to whatever character it is meant to be >.< [13:21:25] idk [13:21:31] I have no idea what \xfe is [13:21:39] where u see it? [13:21:43] hes or something [13:21:47] !part [13:21:52] petan: thats at the end of an article name [13:21:58] in where [13:22:04] sql? http? [13:22:05] hmm, it was parsed from a dum [13:22:07] dump [13:22:10] xml? [13:22:20] just wondering if i can fix the probelm im my list without having to scan all of the dumps again... [13:22:30] hmm [13:22:33] !trout is trouts $1 [13:22:33] Key was added [13:22:39] https://github.com/legoktm/dumpscanner/blob/master/interwiki_finder.py [13:22:44] !trout Cyberpower678 [13:22:44] trouts Cyberpower678 [13:23:10] addshore: http://stackoverflow.com/questions/3027157/what-does-xfe-mean-in-ruby [13:23:32] yus :P but how do i convert a hex exscape char to a char... [13:23:51] there should be a function for that in your language [13:24:01] php, and i just cant find it [13:24:01] xD [13:24:16] currently to fix the unicode bits i have to do [13:24:16] $s = preg_replace( "/\\\\u([A-Fa-f0-9]{4})/", "&#x$1;", $s); [13:24:16] $s = html_entity_decode($s, ENT_COMPAT, 'utf-8'); [13:24:21] http://php.net/manual/en/function.ord.php [13:24:22] which is just plain stupid [13:24:35] Returns the ASCII value of the first character of string. [13:24:44] thats not very usefull :P [13:24:54] !forget trout [13:25:03] !trout del [13:25:04] Successfully removed trout [13:25:11] lol [13:25:22] unless I pregmatch all /\x([0-9][a-z]){2}/ and pass them all through that before re adding them to the strings :/ [13:25:25] you can do something like if (ord(ESC_CHAR) == my_char)) { echo "this is escape!"; } [13:25:26] blergh [13:25:48] but i have strings, not chars [13:25:58] yus :P but how do i convert a hex exscape char to a char... [13:26:03] and the char can be anywhere throughout the string xD [13:26:07] xD [13:26:07] you were talking about chars all the time :P [13:26:13] !trout is /me takes a 10-pound rainbow trout out of a river, and smacks $1 with it over and over. [13:26:13] btw string = array of char [13:26:13] Key was added [13:26:19] !trout Technical_13 [13:26:19] /me takes a 10-pound rainbow trout out of a river, and smacks Technical_13 with it over and over. [13:26:21] and there are likely even articles for these chars that are meant to look like zx34 xD [13:26:23] blergh [13:26:46] heh [13:26:49] Cyberpower678: that doesn't work [13:26:54] petan, how do I program an action command. [13:26:57] wait petan ! http://www.asciitable.com/ [13:27:10] Cyberpower678, zhuyifei1999 can you take this to -offtopic which was created for that [13:27:20] I don't think you can Cyberpower678 [13:27:24] \o/ [13:27:40] if i preg replace \x(([0-9][a-z]){2}) with &#$1!!!!!! [13:27:50] Cyberpower678 that is in my to-do with low priority, action commands are not very useful [13:27:50] petan: I didn't make it here [13:27:56] um I have some problems using public_html on bots [13:28:06] Pyfisch can be describe a bit [13:28:09] * you [13:28:10] 1. thing is I can't create a folder there [13:28:16] mhhhhm that wont work [13:28:25] Pyfisch you should have a folder there [13:28:33] as this is the hex, so in the middle i need to convert the hex to dec [13:28:47] >.< [13:28:52] petan: nice to know that... but I have no [13:28:58] um [13:29:00] that is kind of bug [13:29:03] let me create one for you [13:29:06] Damianz had a script for that [13:29:11] it seems to broke [13:29:46] next thing is, where are the files on the web? [13:29:57] bots.wmflabs.org/~pyfisch [13:31:34] petan: thanks it works [13:31:50] yw [13:42:59] petan: can you please help me check why user local-jimmy can't write to his $HOME/x? [13:43:06] ok [13:44:00] petan: all files I have scp'd to tools, chgrp & chmod'd to the local- user have the same problem [13:44:20] petan: I have to copy these file on tools and nuke the old ones in order to get write permission somehow [13:44:24] yes I see [13:44:31] this is weird [13:44:50] first of all, the permissions on folder are wrong but that shouldn't cause problems [13:45:03] you have a+x but nothing else [13:45:14] which kind of doesn't make sense :) [13:45:20] -rw-rw---- 1 jimmy local-jimmy 0 Jul 12 13:40 x [13:45:27] petan: but stat ~ gives 2771/drwxrws--x [13:45:37] this should allow both jimmy and members of local-jimmy to write there [13:45:42] let me try that [13:46:08] petan: yes that's what I have in mind, but unless I make a copy of the file and set the exact same permission, local-jimmy can't write to it [13:46:58] this is really weird, I think nfs server is broken a bit [13:47:07] for now I recommend you to change owner to local-jimmy [13:47:43] petan: but unless I make a copy of the file, I can't do that [13:47:49] petan: unless I'm root :) [13:47:56] there is a tool called take [13:48:06] you just type "take " [13:48:07] as tool [13:48:13] oh that's convenient [13:49:33] petan: actually my point is to let ~jimmy and ~local-jimmy share the same dotfiles, but chown local-jimmy:wikidev defeats the point by allowing the whole wikidev group writes to them :( [13:50:00] yes I understand, this should be fixed [13:50:12] Coren can do this, I have no powers to do that :/ [13:50:22] I got the powah! [13:50:44] * Coren reads backscroll. [13:51:11] shall I make a new local-unwritable file? :) [13:54:02] jimmyxu: I'm not sure I understand your issue exactly. Can you tell me what you tried to do and how it failed? [13:54:44] Coren: I scp something from my computer to jimmy@tools-login:~/ [13:55:06] Coren: mv it to ~local-jimmy [13:55:12] Coren: chgrp to local-jimmy, chmod to 664 [13:55:19] Coren: and now local-jimmy should but can't write to it [13:56:03] How odd. Can you try it again so I see the exact result? [13:56:11] (without workaround) [13:57:01] Coren: at ~local-jimmy/x [13:57:38] ... okay; that's the expected permissions. Can now, local-jimmy can't write to it? [13:57:46] Coren: $ echo x >> x [13:57:47] -bash: x: Permission denied [13:57:54] Coren: as local-jimmy [13:58:13] Don't touch anything while I look into it. :-) [13:58:17] okay :) [14:03:02] qchris, prod machines are isolated from labs, so going via bastion.wmflabs is definitely not the right choice. [14:03:38] oops, apparently I don't know how to work a scrollbar :( [14:10:09] jimmyxu: Well, I see /what/ is going on. For some reason, that file ends up with acl entries that don't match the file mode. [14:10:26] !screenfix [14:10:26] script /dev/null [14:10:29] Coren: ooh, acl [14:10:47] The interesting question now is "why the hell?" [14:11:29] Coren: same thing that happened to aude? [14:11:46] YuviPanda: No, that file was just created. [14:12:03] Coren: it has something to do with scp? touch x as jimmy then do the same ends up no problem [14:12:05] For some reason, why you chowned or chmodded it, it acquired acl with the original permissions. [14:12:38] jimmyxu: Perhaps your scp client is being too smart for its own good? Could you scp a file like this but not mv or otherwise touch it? [14:13:02] Coren: at ~jimmy/z [14:13:36] Yep. That file ended up with an acl too. That one, however, has group::rw- [14:13:45] Did you copy it any differently? [14:13:54] Coren: nope, did exactly the same [14:14:11] Compare: [14:14:27] -rw-rw-r--+ 1 jimmy local-50568 0 Jul 12 13:56 x [14:14:28] # file: x [14:14:28] # owner: jimmy [14:14:28] # group: local-50568 [14:14:28] user::rw- [14:14:28] group::r-- [14:14:28] mask::rw- [14:14:29] other::r-- [14:14:40] root@labstore3:/srv/tools/project/jimmy# getfacl y [14:14:41] # file: y [14:14:41] # owner: jimmy [14:14:41] # group: local-50568 [14:14:41] user::rw- [14:14:41] group::rw- [14:14:41] mask::rw- [14:14:42] other::r-- [14:15:38] And from what I can tell, you're the only user with acls on files. :-) [14:15:49] Coren: for the record: x was scp'd, and y was created on tools [14:16:19] Coren: but why? I don't recall having done anything related to acls at all [14:16:33] I'm not seeing it either. [14:16:43] Did you create ~jimmy/z on tools? [14:16:48] Coren: nope, scp'd [14:16:58] That one is blissfully acl-free. [14:17:11] Coren: shall I chgrp it again? [14:17:18] Try [14:17:49] jimmy@tools-login:~$ mv z ~local-jimmy [14:17:49] jimmy@tools-login:~$ chmod 664 ~local-jimmy/z [14:17:49] jimmy@tools-login:~$ chgrp local-jimmy ~local-jimmy/z [14:17:50] jimmy@tools-login:~$ become jimmy [14:17:50] local-jimmy@tools-login:~$ echo z >> z [14:17:50] -bash: z: Permission denied [14:17:51] Huh. [14:17:53] Coren: nope [14:17:59] It acquired an acl. [14:18:22] With group::r-- [14:18:33] * Coren ponders. [14:19:59] Aha. I can reproduce the problem if I touch in a home and mv to a project. [14:20:38] touching directly in the project dir from the normal user creates the file without acl and with the expected permissions. [14:21:13] Coren: but I touched 'y' in ~jimmy and then moved it [14:21:19] Coren: ends up writable [14:21:47] jimmyxu: Still has an acl. The difference is that when you created the file, it had g+w so the acl was also created with g+w. [14:22:10] * jimmyxu checks umask [14:22:13] The difference in that case is that the acl "happens" to be right. [14:22:29] when bots is closed finally, will there be new instances on tools? [14:22:46] Pyfisch: We're adding exec nodes as load demands. [14:23:19] Coren: yea.. umask 022 then the problem appears [14:23:32] jimmyxu: /Something/ is trying to be too smart about preserving permissions at all costs. :-) [14:25:04] The problem is that I didn't turn on ACLs at the nfs level, so that from the client you can't even /see/ that the file has 'em. [14:25:27] No, actually, the problem is that it acquires them. It's made more obscure by not being able to see them. [14:26:02] so it becomes magic [14:26:30] * Coren tries to figure out why it happens and why it /apparently/ never happened before. [14:27:04] Can you try to scp something straight to ~local-jimmy? [14:27:15] (i.e.: not to your home then mv) [14:28:38] Coren: ~local-jimmy/f [14:28:58] Ah, see, /that/ one doesn't have stealth acls added to it. [14:29:20] So it's apparently the move accross filesystems that's trying to do something funky to preserve permissions. [14:30:38] * jimmyxu apparently is too lazy to scp into local- [14:30:56] needs to type the path.. won't work [14:31:35] Yeah, that'd allow you to work around the problem, but it's still annoying that it exists in the first place. [14:32:53] jimmyxu: Can you do the scp-mv things again see if you see a difference? [14:33:50] Coren: seems it worked \o/ [14:34:39] I forcibly remounted the underlying filesystem noacl, so whatever trickery was going on couldn't work anymore. That does mean I can't turn on /actual/ acl support for users though. [14:35:00] Which is a shame. [14:36:57] * Coren will need to research that odd behaviour. Perhaps the NFS server notices that both nfs mounts are on the same filesystem and tries to be too smart for its own good, or there is something odd with the way the kernel does a rename() on NFS when the fsid of source and destination match. [14:37:29] * jimmyxu hands Coren some cookies :) [14:39:20] ... [14:40:19] It's definitely a bug though. Even if I had ACLs turn on all the way and you could have seen it, I can think of no reasonable semantics that would allow mv to quietly add an acl. [14:42:01] Coren, have you seen #wikimedia-labs-offtopic [14:43:27] ... [14:43:34] Technical_13: ... yes? [14:44:13] Cyberpower678: That seems to be... unneeded. If it's so offtopic that it belongs on -labs, certainly any other channel would do and there is no need for a dedicated channel? :-) [14:44:39] (+1 to Coren) [14:44:39] Coren, petan made it. [14:45:16] Cyberpower678: It'd seem unneeded even if the potus had made it. :-) [14:45:30] I just go with it. [14:45:32] :p [14:51:08] ... [14:55:17] Why am I still @q in offtopic? [15:11:35] petan, you'll find an interesting log for the offtopic channel you created. [16:03:35] Coren: Re the email on labs-l, there's no need for /etc/labs since there's already three other files in /etc that should serve [16:04:43] anomie: well, there's /etc/wmflabs-instancename whose presence can be tested for, but I would have liked to add /etc/wmflab-project so that you can (for instance) check between "tools" and "toolsbeta" [16:05:46] (without having to parse and guess from the instance name) [16:06:06] I agree that /etc/wmflabs-project could be useful for some purposes, not necessarily related to that email discussion [16:06:34] Well yeah, regardless of what I did I would have wanted it to be of general applicability. :-) [16:14:51] I'm not getting his even worriying about the specific cluster numbering though. In which case would code caring about which shard a specific database is on not be a bug? (Or at the very least, an unhealthy environmental dependency) [16:16:00] I can't think of any, except *maybe* for knowing if you can join cawiki against eswiki or something weird like that. [16:17:21] Even then, that'd be a bug; you can't rely on any two databases being on the same cluser any more than you can either of them being on a specific one. A tool that did joins like that would simply break randomly. [16:18:14] Speaking of cross-database joining, I see dewiki_p needs to join against wikidatawiki_p while everything else needs to join against wikidatawiki_f_p? [16:19:14] Break randomly, or only offer random cross-wiki join options based on the current situation. Either way, not too useful. [16:23:06] anomie: Hm. Good point. I should provide the federated version even on the DB's home shard. [16:23:18] For much the same reason. :-) [16:24:05] (Although the really /smart/ code would check for availability of the unfederated version and use that one if it is available since performance will invariably be better) [16:25:03] * Coren wonders if mysql allows table aliases. [16:25:12] s/table/database/ [16:25:48] Alternately, I can just make _f_p on home databases to be another _p [16:27:22] petan: Someone's hope dir doesn't exist [16:28:11] They have no hope? :-) [16:28:24] Homedir gets created on first login. [16:28:26] loss of hope [16:28:30] Coren: yeah - it changed [16:28:45] Coren: while I understand the maintenance issues related to shards and databases, a couple of years ago I did some research work that benefited greatly from knowing which Wikipedia editions were found on specific shard servers [16:29:18] Coren: conventient, fast, and easy to have the database server do the join rather than doing it manually for those cases where both languages I examined where on the same server [16:29:26] Nettrom: But that's a really marginal application. It'd benefit from, say, a table having the mapping handy but not from being able to connect to shards by number. [16:29:53] petan: Fixed [16:30:20] Laptop track pads such when you're sweaty :( [16:30:26] Coren: yeah, as long as I can figure out what other Wikipedia editions were available from a given server, it would work... and yes, it's a marginal application, just thought I'd throw it out there as an example where it was really convenient to know [16:30:56] Nettrom: Any application connecting to sX and expecting to find some database there has an unreliable environmental dependency. [16:31:11] If someone really wants to know if two wikis are on the same shard, they could always just connect to the host for the first and see if the database for the second exists too. Or probably check if the dns entries for both resolve to the same IP. [16:31:44] Yeah, seeing if the names point to the same place is guaranteed to be accurate. [16:32:02] (and probably the simplest) [16:32:05] yeah, that would work too, good call [16:33:53] * Coren waves to Platonides since he's the one who prompted the discussion in the first place. :-) [16:44:56] Coren: To see the max mem a job ended up using I do qstat -j , right? And then maxvmem ? I ran the first visualeditor diffbot job manually just now with -mem 2048m to see how much it needs. [16:45:18] Krinkle: Yep. That works. [16:45:37] Can I see that when the job has finished? [16:45:56] You can also use qacct after the fact (there is a ~5 min lag for the stats to propagate) [16:45:56] ok [17:00:10] [bz] (8NEW - created by: 2Chris McMahon, priority: 4Unprioritized - 6major) [Bug 50622] Special:NewPagesFeed intermittently fails on beta cluster; causes test failure - https://bugzilla.wikimedia.org/show_bug.cgi?id=50622 [17:41:20] [bz] (8NEW - created by: 2Chris McMahon, priority: 4High - 6enhancement) [Bug 50335] support dvwiki in beta labs - https://bugzilla.wikimedia.org/show_bug.cgi?id=50335 [17:53:46] [bz] (8NEW - created by: 2Tim Landscheidt, priority: 4Normal - 6normal) [Bug 48626] Provide wiki metadata in the databases similar to toolserver.wiki - https://bugzilla.wikimedia.org/show_bug.cgi?id=48626 [18:06:23] [bz] (8NEW - created by: 2Quim Gil, priority: 4Normal - 6enhancement) [Bug 51050] Connecting wikitech.wikimedia.org user profiles with community metrics - https://bugzilla.wikimedia.org/show_bug.cgi?id=51050 [18:09:21] yes, i am a bot [18:42:40] Coren: hah, jobs that require CPU! [18:43:15] No, his implicit question was sane; it's just that requesting cluser access in our setup would be pointless. :-) [18:44:41] ok... i didn't read the whole thread [18:44:56] i still think you were funny though :) [18:53:18] You don't need no CPU [19:27:25] I have a black belt in malicious-free snark. :-) [19:30:26] ;-) [19:35:23] Coren: https://wikitech.wikimedia.org/wiki/New_Project_Request/DOI_Bot [19:35:45] I posted on the talk page and marked the request completed, but you may want to talk with the user directly [19:35:53] I'll gently deflect him towards tools. [19:35:58] * Ryan_Lane nods [19:36:00] Ah, I see you already did. Will do. [19:37:04] I should say that the new project request form should probably start with a blurb along the lines of "You almost certainly don't need a new project..." and some guidelines for when it's apropriate. Do you mind if I do so? [19:37:25] go for it [19:37:25] don [19:37:25] err [19:37:33] don't make it seem like people should only ever use tools, though :) [19:37:56] though 90% of the time that'll be the case [19:42:55] I'm thinking something along the lines of "For most bots and self-contained web services that do not require direct system administration (point at tools); some existing project might already exist that fits the bill (point at list of projects)" [19:44:31] * Coren tries to find the right spot for this. Editnotice? [19:45:49] I think there was a 'request new project' page? [19:45:56] that would be the proper place [19:47:43] It's an edit-by-form. [19:51:57] Ryan_Lane: Do we have a canonical list of projects we can point people ''at''? [19:59:45] Ryan_Lane: https://wikitech.wikimedia.org/wiki/MediaWiki:Editnotice-0-New_Project_Request [20:02:09] Cyberpower678: you here? [20:19:26] hah, the devil's advocate [20:19:31] Platonides: :-) [20:22:13] Coren: https://wikitech.wikimedia.org/wiki/Special:Ask/-5B-5BResource-20Type::project-5D-5D/-3F/-3FMember/-3FDescription/mainlabel%3D-2D/searchlabel%3Dprojects/offset%3D0 [20:22:30] maybe without the member list [20:22:40] one sec [20:22:55] Ryan_Lane: Yeah, already found it; but turns out FormEdit doesn't like editnotices; Ima change the form instead. [20:23:09] https://wikitech.wikimedia.org/w/index.php?title=Special:Ask&q=%5B%5BResource+Type%3A%3Aproject%5D%5D&p=format%3Dbroadtable%2Flink%3Dall%2Fheaders%3Dshow%2Fmainlabel%3D-2D%2Fsearchlabel%3Dprojects%2Fclass%3Dsortable-20wikitable-20smwtable&po=%3F%0A%3FDescription%0A&eq=no [20:23:18] how is that query *longer*? [20:25:33] Coren: https://wikitech.wikimedia.org/wiki/Special:Ask/-5B-5BResource-20Type::project-5D-5D/-3F/-3FDescription/mainlabel=-2D/searchlabel=projects/offset=0 [20:25:36] there we go [20:25:38] stupid SMW [20:28:03] better: {{#ask: [[Resource Type::project]]|?=|?Description|limit=0|mainlabel=-|searchlabel=list of projects}} :-) [20:28:57] Ryan_Lane: https://wikitech.wikimedia.org/wiki/Special:FormEdit/New_Project_Request [20:29:04] Iz the pretties. [20:29:14] indeed [20:29:21] much better :) [20:29:54] Incidentally, perhaps removing the "request completed" checkmark would be good. I can't think of a good reason for it to be there on creation. :-) [20:30:27] hm. I did that somehow for another page [20:31:02] I think I made one form for creation and another for editing [20:31:08] You did. [20:31:18] So we should be able to remove the checkmark from the creation one. [20:31:52] yep [20:32:43] done [20:34:37] Hi! Does anyone have a moment to help me with my Tool Labs account request? This is kirsten, btw. Just want to check it out for the docs. [20:35:24] Ryan_Lane: Look! Useability! https://wikitech.wikimedia.org/wiki/Special:FormEdit/New_Project_Request [20:36:17] (You can't just remove the field, I think, I've added it back as hidden with default No) [20:36:33] kma500: Sure, Give me a second [20:36:46] thank you! [20:36:58] Coren: oh. you can't remove it? [20:37:11] ah. right. makes sense. it needs to have a value [20:37:26] Ryan_Lane: I think if it's not in the form at all, it wont end up on the resulting page (rather than there, but with "No") [20:37:36] yep [20:37:59] I wish bugzilla accounts were linked with wikitech accounts [20:38:17] kma500: {{done}} [20:38:26] great! Much appreciated. [20:38:44] zzzz dabase queries are slow [20:39:32] OrenBochman: No they're not. [20:40:12] Well, they can be - in absolute terms - but they are blindingly fast in their longness. :-) [20:40:19] What are you trying to do? [20:41:09] I'm trying to group revisions in sessions [20:41:22] in-> into [20:41:34] (for a given user [20:41:35] If you want me to take a look at your queries to see if they can be made faster, don't hesitate to ask. [20:41:53] I'd like to see if this one works [20:42:04] As defined by... interval between edits? [20:42:17] 600 seconds [20:42:48] (idealy an hour) [20:42:48] * Coren hopes you're using revision_userindex and not revision :-) [20:43:00] I am not [20:43:14] * OrenBochman kills the query [20:43:16] Well, /there's/ your problem! [20:43:43] If you have a where clause on user ids or names, you *really* want to use the revision_userindex view. :-) [20:44:43] is that documented anywhere ? [20:46:02] ... I don't think it is? [20:46:20] hmmm [20:46:47] And, indeed, it really should be. [20:47:11] We have a documentation sprint coming soon. I'll add this to the do-not-forget. [20:48:12] (There's a couple such tables) [20:49:59] good [20:50:19] couase running queries on revision takes a long time ;-) [20:52:04] * YuviPanda hopes we have HDFS / Hadoop for this within the next... 2 years?  [20:52:10] should make running these kinda queries much nicer [20:52:34] ;-) [20:54:19] perhaps once hadoop 2.0 comes around [20:56:31] hmm faster [20:56:49] I'll try the self join now [21:01:28] kind of slow [21:02:00] take 2.5 minutes to process get my first 4 edits [21:06:43] Technical_13, I am now. [21:07:07] Had some questions about cyberbot.. [21:07:51] Is there anyway to configure cyberbot to ping people using the template when there are new nominations? [21:08:57] is there a mysqlfiddle of the db anywhere ? [21:26:58] I guess it takes 2.5 minuts to process everything too [21:30:04] OrenBochman: Honestly, given your problem, I'm not sure the DB is the right place for it at all. IMO, it'd be much faster to simply fetch a sorted list of all of a users' revisions and build the session list in-app. [21:30:48] And bandwidth isn't an issue since I'd expect the only thing you'd want is to select just the timestamps. [21:34:35] Cyberpower678: Is there anyway to configure cyberbot to ping people using the template when there are new nominations? [21:34:58] Technical_13, what? [21:35:15] Template? Nominations? Ping who? [21:35:19] Cyberbot is your bot, right? [21:35:29] mmm hmm [21:35:44] The RfA/B status bot? [21:36:32] Who's it supposed to ping? [21:36:58] Whoever uses the template that opts in? [21:37:15] Was something another user was asking me about... [21:37:42] A js script would suffice. [21:39:31] Cyberpower678, Hi [21:39:48] tos, hi [21:40:06] T13 asked me to come here to talk to you [21:40:15] tos, about what? [21:40:40] I was wondering if there is a system in place/that can be built to ping users whenever a new RfA begins [21:40:57] Quite a few users would opt in for such a system, I suppose [21:40:57] tos, ping them how? [21:41:12] Echo would be one of the options [21:41:21] A bot would be another [21:41:32] Cyberbot does not yet support Echo [21:41:35] And leaving messages to the users directly/mailing them would be more [21:41:52] Cyberpower678, But there can be bots which could, right? [21:47:06] tos, I can easily program echo into Cyberbot's supporting framework, but why couldn't someone create a js script to be installed into the user account. [21:47:32] Cyberpower678, I have no idea why not. [21:47:39] I'm just looking for options here [21:47:58] As I said, I just am suggesting this so someone can help implement it :) [21:48:01] That js can monitor that template and create an OBoD. [21:48:49] An echo ping will be a lot better I think [21:49:07] Because thats the standard for most such communication now. [21:50:19] tos, after further thought. There's nothing out there to let me tie into Echo. [21:51:18] I was hoping there would be some similar functionality option available for echo [21:51:59] A primitive way to ping would be to simply make talk page messages to self, linking the userpages of all those who have opted in [21:52:08] Cyberpower678: wouldn't the logical answer be for tos to use the template on his talk page so he'll get a ping every time it is updated? [21:52:43] The only way Echo can work is to have cyberbot place links everywhere mentioning the users or to place a talk page notice. [21:52:43] Something like [[User:Abc]],[[User:Def]],[[User:ghi]] Xyz has started his RfA ~~~~ [21:53:02] That was among the options I considered [21:53:26] The results would be something like an echo notice saying Cyberbot I mentioned you on User:Cyberpower678/RfX Report [21:53:34] Which would say nothing. [21:53:49] And a talk page notice is too intrusive. [21:54:13] a Gadget would be simple enough [21:54:39] and can even 'fake' its way into echo [21:54:49] Technical_13, no. [21:54:54] How will that ping him? [21:54:57] but would be slowing the user down a little bit, I think [21:55:22] YuviPanda, Infinitesimally small slowing, I think. [21:55:46] tos: 1 (at least) network request every x [21:55:53] YuviPanda, interesting. How can you generate an echo notification with a js gadget? [21:56:11] Cyberpower678: it'll be a major hack, but you'll essentially find the Echo DOM elements and insert it into it :D [21:56:20] there is nothing you can't do, really [21:56:20] * tos sits back and hopes the three of you would figure it out. [21:56:29] That sounds like a major slowdown. [21:56:38] Cyberpower678: how so? [21:56:49] Cyberpower678: the network request? yes (not major, but). But DOM manipulation? [21:56:50] no [21:56:53] Because hacks usually do. [21:57:04] well, this is an ugly hack, not a slow one. [21:57:13] DOM manipulation is the same speed, no matter how hacky or clean it is [21:57:50] YuviPanda, if we're going to notify people, we're doing it without hacking, slowing down, or sneezing. [21:58:06] DOM won't slow anything down, no. [21:58:09] the network call? yes. [21:58:09] It's going to be pretty and unobtrusive. [21:58:10] Cyberpower678: everyone gets a notification everytime there is a post to their talk page.. it is the only notification that can't be turned off. [21:58:34] if you want the *code* to be pretty and unobtrusive, then yeah, Technical_13's is the only way to go [21:58:49] Technical_13, I still fail to see how translcuded the RfX reporter will ping the user. [21:59:13] It is a talk page post. [21:59:27] Technical_13, not really. [21:59:35] At its very core, that is all it is. [21:59:51] When the reporter updates, the contents of the talk page will be exactly the same as before. [22:00:28] I see. So the bot only updates the template and not each instance of the template. [22:00:53] (take page crap){{User:Cyberpower678/RfX Reporter}}(more talk page crap) [22:01:06] RfX updates. Talk page will still be: (take page crap){{User:Cyberpower678/RfX Reporter}}(more talk page crap) [22:01:14] Cyberpower678: what if the talk page modified its contents based on which page it is at? [22:01:14] Nothing is updated [22:01:24] hmm [22:01:32] you are right, that wouldn't change anything [22:02:00] * Cyberpower678 doesn't like hacking. [22:02:15] * Cyberpower678 scans the API. [22:02:34] * YuviPanda agrees with Cyberpower678, goes off to do something else [22:02:35] * tos hands an axe to Cyberpower678. [22:02:58] * Cyberpower678 doesn't know what to do with the axe. [22:03:01] Then... the option that would be required would be to build a subscriber list on user talk:cyberpower678/rfx reporter [22:03:10] So [22:03:41] Cyberpower678, Hacking. [22:03:42] Obviously, but that still doesn't answer how we are going to notify these people. [22:03:56] Technical_13: can you help me test that? [22:04:00] Technical_13: what's your enwiki username? [22:04:21] .so.. just parse the api for all uses of the template and build a user list and sign it and it would use notifications for the mention unless they have that feature off. [22:04:31] Technical_13 [22:05:13] What you testing? Lol [22:05:31] Oh.. [22:05:49] "Cyberbot I mentioned you on User talk:Cyberpower678/RfX Reporter [22:05:50] tos: I know what you can do... [22:06:04] Technical_13, What? [22:06:05] Add the template to your watchlist. [22:06:19] Then you'll see when it is updated. [22:06:32] Except it would be sharing my watchlist with 300+ more pages [22:06:37] Email is another option. [22:06:45] You can also add it to your wm-bot recentchanges feed on IRC [22:06:59] Technical_13, don't do that. [22:07:18] You'll get pinged every hour at least. [22:07:43] And if it's an active RfX, every minute. [22:07:47] or second. [22:07:58] Cyberpower678, Unless the bot can be made to only do it on the creation of a page [22:08:01] The reporter now updates the moment it sees a change. [22:08:48] Cyberpower678: I talked to the Echo devs, apparently an API is 'deprioritized', not happening anytime soon [22:09:02] meh, this makes Echo fairly useless for gadgets / bots /tools [22:09:48] Well that makes Echo out of the question. [22:09:52] yeah [22:10:15] * tos pretends to understand everything discussed and nods [22:10:21] Email or a talk page notice is the only remaining option then. [22:11:00] I'd say email is more attention getting. [22:11:50] Cyberpower678, Why not an echo ping the way I said it? [22:12:03] Because the API doesn't support it. [22:12:17] Technical_13: did you get a notification? [22:12:20] about a mention? [22:12:20] Not that way [22:12:26] The mention way [22:12:30] Technical_13: on two pages? [22:12:39] Cyberbot I mentioned you on User talk:Cyberpower678/RfX Reporter [22:12:48] doesn't really say much. [22:12:54] Anyone else having tools-login problems? [22:13:24] Cyberpower678, Unless the edit summary has a link to the RfA [22:13:41] fwilson, yes. [22:13:47] I was making supper.. I'll check [22:13:56] Looks like load is on the rise. [22:13:58] !mytalk Technical_13 [22:14:02] ...wonderful [22:14:13] ... helpmebot isn't here... [22:14:26] heh. shit. I just realized I may be spamming people with emails from my development vagrant instance [22:15:32] fwilson, and we're online again. [22:15:51] Cyberpower678: excellent, thanks to whoever fixed it :) [22:15:52] Guys? [22:16:01] What can we do about the pings? [22:16:11] fwilson, Probably just a temporary spike. [22:16:18] It happens from time to time. [22:16:36] Cyberpower678: alright [22:16:56] YuviPanda: nothing [22:17:05] Technical_13: on enwiki? no ping at all? [22:17:20] None [22:20:15] Technical_13: meh, echo is not usable for this at all then. I mentioned you on my talk page... [22:20:51] You sign it? [22:20:56] ~~~~ [22:21:40] Technical_13: no, it was a template [22:22:02] Technical_13: now? [22:22:20] Technical_13: now? [22:22:31] That is why. Echo needs [[User:Technical_13]] and ~~~~ to ring echo [22:24:43] Technical_13: oh, it does text parsing? [22:24:49] and not DOM parsing? [22:25:47] Not entirely sure how it does it, just that it needs those elements to work. [22:26:25] Will even work with user link display:none; [22:26:56] hmm [22:27:09] Technical_13: so my current test failed and you got no pings? [22:27:28] http://enwp.org/Template:Pinggroup works without displaying names hidden. Correct. Nothing from your tests. [22:29:59] hmm [22:33:11] hi again. I can't ssh to my tool labs account, and I think it's because I put the wrong shell account name when I created my wikitech account. I can't figure out how to change that. Can anyone help? [22:39:42] kma500: You can't actually change your shell account name, I'm afraid. [22:40:20] kma500: It is kmenger [22:42:17] Hm. My mac account is kirsten. Do you have advice? Not sure what's best/easiest. [22:52:27] kma500: ssh kma@tools-login.wmflabs.org [22:52:40] or ssh -l kma tools-login.wmflabs.org [22:55:25] Ryan--still getting permission denied. [22:56:55] kma500: did you upload your public ssh key to wikitech? [22:57:03] yes [22:57:29] it's showing up, at any rate. [22:57:37] in the ui [22:58:50] kma500: what's your wikitech username? [22:58:56] Kmenger [22:59:15] ah [22:59:20] your ssh username is kmenger [22:59:37] so: ssh kmenger@tools-login.wmflabs.org [23:00:08] that works! [23:00:41] thank you [23:04:26] yw [23:04:40] [bz] (8NEW - created by: 2Peter Bena, priority: 4High - 6critical) [Bug 49811] Block user is broken on wikitech - https://bugzilla.wikimedia.org/show_bug.cgi?id=49811 [23:58:11] evening [23:58:17] evening [23:58:54] So, I'd like to do some experimenting on a labs machine. Ryan made me an account more than a year ago, but I don't remember how it goes at all. Where's the quick-start page? [23:59:25] Let me guide you. First, do you have shell access?