[00:42:06] Is there a stream on the labs that includes the text of revisions? [00:52:41] a930913: I don't think so. [00:53:28] scfc_de: Why isn't that a basic provision? :/ [01:00:44] a930913: I would be happy if we already had a copy of the Redis queue for just the RC metadata :-). [01:04:45] scfc_de: But why don't we? :/ [01:11:17] a930913: https://bugzilla.wikimedia.org/show_bug.cgi?id=59721 if you want to take this on. [04:44:55] hello. [04:45:41] I am not able to connect to http://tools.wmflabs.org/tusc/tusc.php from my bot, but I am able to connect to it from my local machine. this is a problem because I cannot verify anyone's password. Could someone look into this? [04:46:05] if you type wget http://tools.wmflabs.org/tusc/tusc.php into the command line, it will timeout [04:54:26] Magog_the_Ogre: tools.wmflabs.org resolves to an "outside" IP, and you cannot reach these from inside Labs. You need to access http://tools-webproxy/tusc/tusc.php instead (tools-webproxy being the Labs host that incoming traffic for tools.wmflabs.org is routed to). [04:55:50] thanks [11:40:37] hmpf. globalblock on deploymentwiki seems broken O_O [13:01:49] Hi, how can I connect from labs to a other labs tool? In my case connect to http://tools.wmflabs.org/tusc/tusc.php from my tool. if I ping "tools.wmflabs.org" from the command line, dns resolves it with the ip 208.80.155.131 and then I get a timeout. [13:03:13] Luxo: try replace 'tools.wmflabs.org' with 'tools-webproxy' [13:03:28] ok [13:07:52] hedonil: thanks, it works [13:09:37] Luxo: yw. Coren will try a hack to keep url http://tools.wmflabs.org working when called form within labs. For now tools-webproxy will do the trick. [13:10:21] ok [13:43:36] hi. I am having problems to ssh to bastion.wmflabs.org, while I can connect to bastion-eqiad.wmflabs.org [13:43:43] can anyone help? [13:56:25] Mpaa: works for me. Any details? [14:00:55] hedonil, on bastion I get: [14:00:57] debug1: Offering RSA public key: /home/squiddy/.ssh/id_rsa [14:01:09] debug1: Next authentication method: publickey [14:01:30] on bastion-eqied, the key is accepted [14:01:55] debug1: Server accepts key: pkalg ssh-rsa blen 279 [14:04:03] Mpaa: Any changes whith the keys in last couple of days? maybe not properly synced. [14:05:06] hedonil, I generated the new key a few days ago. But I do not understand why the same key is OK on bastion-eqiad and not on bastion [14:05:24] some kind of chache?? [14:05:32] cache?? [14:06:06] because from another computer I can access bastion with same user and a different key [14:07:13] Mpaa: ther's a process syncing the keys from https://wikitech.wikimedia.org/wiki/Special:Preferences, maybe didn't work properly. [14:09:04] Mpaa: don't know if that'll help, but try bastion2.wmflabs.org. another bastion in pmtpa. [14:09:33] hedonil, I tries to insert/remove the key a few times in te past days, no success; tried also bastion2 no success [14:10:26] the key should be OK as on bastion-eqied is OK [14:10:36] Mpaa: then we'll have to poke Coren or scfc_de to take a deeper look [14:12:22] hedonil, thanks, hope Coren os scfc-de will look into it [15:45:47] Coren: got a moment? or maybe be able to point me in the right direction? [15:46:15] If it's really quick, I'm on my way out. What can I do to you? [15:46:45] Need directions for contacts on how to get UTRS back up from the "suspended" status [15:47:04] with the migration [15:48:26] andrewbogott_afk is almost certainly the one to ask, but chances are that you can simply reboot the instance and it will come up with only minor issues. [15:48:42] That's true for most (though not all) automated migrations. [15:48:57] k, i'll try and reboot the puppy then [15:49:23] You may also want to review the security groups and see if they still make sense. [15:49:35] yep [16:04:24] is it tool late to run finish-migration ? [16:08:29] johang: No; in fact you probably will have to. [16:08:52] Coren: do I do it from new site or old site? [16:09:25] johang: New site; there isn't an old site you can connect to anymore anyways. [16:10:48] Coren: thanks. running it now. [16:14:30] Coren: Hehe. Already thought about expanding the storage volumes? [16:14:40] Coren: got the database but it didn't restore the crontab [16:14:47] used: 45% /data/project [16:15:00] used: 88% /public/dumps [16:16:33] andrewbogott_afk: i'll be around until your around, so pls ping me when your back [16:16:37] hedonil: /data/project contains redundant copies for the moment. /public/dumps is pretty much at objective; we simply adjust how long we keep. [16:16:38] screen is not working after running "become ". known issue? [16:17:01] MOAR! [16:17:27] ;) [16:21:43] johang: It's not so much an issue as "the way screen works". [16:23:43] Coren: crontab and access.log disappeared in the migration. no big deal though. [16:25:46] johang: By design. As explained in the migration emails, logfiles were not copied over by default. Your crontab should be there though. Is there a file names ...DATA.crontab in your tool's home? [16:26:39] Coren: only ...DATA.olduser [16:26:56] not sure which of the old machines the crontab was on [16:32:01] hi! I'm wondering about the my tools database [16:32:25] where is it? (new username: "s51171") [16:33:31] previously, I had placed it on the same db server as the svwiktionary database (because I want to be able to join), but I can't find it any more [16:36:05] andrewbogott_afk: actually could you drop me an email, deltaquadwiki@gmail.com because I do have to go, but i'll respond fast with that [16:41:12] skalman12: It is still there, with its previous name (your new credentials should have full rights to it) [16:41:55] Coren: how do I know my old name? previously I just read it from replica.my.cnf [16:43:07] Check your home, there should be a file named ...DATA.olduser containing your previous username. [16:43:14] Coren: found it. [16:43:36] Coren: so am I supposed to continue with the old name? [16:45:09] skalman12: You can just use the database with its old name. Alternately, you can dump it, drop it, and restore it with your new name. Either way's good. [16:45:37] how do you dump it? [16:45:42] mysqldump [16:45:56] oh, ok [16:46:14] But there's no /need/ to do it if you're okay with the old name. [20:25:01] andrewbogott: FYI: [15:46:45] Need directions for contacts on how to get UTRS back up from the "suspended" status ff. [20:25:19] hedonil: I just fixed it, I think. [20:25:29] andrewbogott: ;) [21:13:50] andrewbogott: Adding a service group as a member to a service group (cf. https://wikitech.wikimedia.org/w/index.php?title=Special:NovaServiceGroup&action=managemembers&projectname=tools&servicegroupname=local-jarry-common&returnto=Special%3ANovaServiceGroup) adds "uid=local-wikicup,ou=people,ou=servicegroups,dc=wikimedia,dc=org" (NB: uid=local-, but ou=servicegroups). I suspect this is a bug? [21:14:55] scfc_de: That looks right to me, I think...? [21:15:12] service group names start with local- by default, in tools at least. [21:16:32] You're still in last year AFAIUI :-). We switched to $PROJECT.$TOOL, i. e. cn=local-jarry-common,ou=groups,cn=tools,ou=projects,dc=wikimedia,dc=org became cn=tools.jarry-common,ou=servicegroups,dc=wikimedia,dc=org. [21:17:29] Ah, ok, that's right. Hm... [21:17:45] The behavior is different with a nested service group than with a top-level one? [21:21:28] Hmmm? I think (haven't tested) that the problem is (will be) that the purpose of nested service groups is that a service group A has access to the files & Co. of another service group B, and if in the member list of B the name of A is mangled, it doesn't get that access. [21:21:54] *A doesn't get that access. [21:22:39] OK, so the new service group isn't itself misnamed, it's just the entry unter the other group that's wrong? [21:25:48] Yes (I think). If the service group wikicup is added as a member to the service group jarry-common, the member entry in new-LDAP should be tools.wikicup,ou=people,ou=servicegroups,dc=wikimedia,dc=org (I think). (We still keep both OUs of LDAPs updated I believe?) [21:27:51] andrewbogott: So should I file a bug? [21:28:11] Hm, yes please, I don't think I can look at it just now. [21:29:07] andrewbogott: Okay. Another short question: A wikitech user asked if it is possible to rename his wiki username. He also tried creating a new wikitech account, but he can't use the shell account name he chose for his first wikitech account again. Any good advice? [21:29:25] scfc_de: Why is the IRC RC feed so hard to parse? D: [21:30:22] scfc_de: For the moment, the advice is "You can't change your wiki username" [21:30:49] andrewbogott: That's what I thought :-). Thanks! [21:30:58] Changing things would take a fair amount of digging around. And reusing an existing shell name is a potential security risk. [21:31:10] a930913: Aren't there libraries for that or so? [21:32:13] andrewbogott: His "old" account is a day old, so the risk would be manageable. [21:34:12] a930913: I am not opposed to anything that is better than current IRC feed, just there isn't anything better proposed so far :P [21:34:16] andrewbogott: But on the other hand a "different" shell name should be a minor inconvenience. [21:34:24] everything else was either worse or did suck the same [21:34:51] well, I once proposed something that would be better but people from ops said it would be too hard for them to maintain it [21:35:06] so either it's developer friendly or operator friendly [21:37:12] omg [21:37:33] well, that is just stupid [21:38:33] here we go [21:38:41] * petan goes fix his irc client [21:38:48] scfc_de: There are? Where? [21:38:55] a930913 [21:38:59] I don't believe that [21:39:07] there are no libs for parsing wikimedia irc feed :P [21:40:26] a930913: I don't know. But if there are x tools parsing that feed, a spirit of collaboration would produce them, not? :-) [21:41:15] petan: Is there any documentation on what the feed is? [21:41:33] I don't know but there are open source tools that use it [21:41:44] it's also pretty easy to understand that just by reading it [21:41:45] scfc_de: I've got stuff that parses /most/ of it, but not /all/ of it. [21:42:00] basically it's just some weird list of info separated with IRC symbols for colors etc [21:42:29] a930913: pywikibot.botirc [21:42:53] (if you're looking for a python-based lib) [21:43:08] not sure how well it works though - I just remembered there was something in pywikibot and grepped for it [21:44:42] i do have a code snippet for that [21:45:25] a930913: what do you mean by most? [21:45:58] gifti: Most edits I suppose. [21:46:38] so, what's missing? [21:46:57] gifti: Various logs, af hits, etc. [21:47:18] valhallasw: Found the regex it uses. I'll try it, but I don't think it'll work. [21:49:26] af? [21:53:22] gifti: Abuse filter. [21:54:55] ah [21:59:15] valhallasw: Yeah, doesn't work with deletes, protects, newusers, moves and so on. [22:01:52] this works for me: https://tools.wmflabs.org/giftbot/irc.tcl [22:16:39] gifti: Can you help me translate the regex into python? [22:16:57] uh [22:20:02] gifti: come on! ;) http://www.flickr.com/photos/110698835@N04/11228026156/in/photolist-i7byZS [22:21:10] fu! [22:21:22] * hedonil laughs [22:23:37] a930913: where/what is the python regex you found? [22:24:56] gifti: Line 63 of irc.tcl. [22:25:41] haha https://www.youtube.com/watch?v=4XpnKHJAok8 this video has nice intro [22:26:03] explanation of why users of git feel like idiots XD [22:29:19] Talking of idiots, https://github.com/torvalds/linux/pull/82 is hilarious [22:37:15] !rofl is https://github.com/Feminist-Software-Foundation/C-plus-Equality [22:37:16] Key was added [22:39:56] var int privilege "Yes"; [22:39:57] // this may look wrong, but the int is allowed to FEEL like a String. [22:40:05] :D [22:45:21] a930913: i meant botirc.py, but nvm [22:45:45] it seems to be basically the same with just some subtle differences [22:49:01] key could be that the closing parenthesis in the size part isn't optional [22:49:20] iirc it is omitted in log entries [22:53:37] * hedonil apologizes for beeing mean to gifti and hands over a bouquet of flowers ❀⚘❀⚘❀⚘❀⚘❀⚘❀⚘❀⚘❀⚘ [23:00:22] gifti: I've tested the regexes, and neither work. [23:01:11] so the tcl and the python one or the python and the corrected python one? [23:03:10] gifti: Any regex tbh. [23:03:20] hm [23:03:28] maybe encoding issues? [23:04:08] gifti: There are a number of formats used, that's the problem. [23:04:21] no, actually just one [23:04:37] sometimes cut off by irc line limit [23:05:18] gifti: Yes, but that's another issue :| [23:06:59] gifti: If there was one format, why wouldn't the existing regexes work? [23:07:13] what channel are you listening to? [23:07:31] #en.wikipedia [23:17:25] a930913: they could be malformed [23:17:38] i'm testing mine atm, so far no misses [23:18:30] gifti: AF hits? [23:18:35] yes [23:18:37] plenty [23:20:46] gifti: Maybe it's lots in translation? [23:21:02] which translation? [23:21:08] From tcl. [23:21:22] i'm testing it in tcl, so there is none ;) [23:21:58] did you change \) to \)? [23:22:04] gifti: As I said, when I've translated it, it's not working perfectly. [23:22:43] so af is never recognized? [23:28:33] gifti: Not sure if ever, but some certainly are. Hold on, I'll show you. [23:29:45] gifti: http://i.imgur.com/AAx0Ghl.png [23:36:52] when was that? [23:37:58] gifti: Erm, ages ago. But there was nothing special about that one, there were loads. [23:38:27] hm, ok, just wanted to check if i catched it [23:39:09] gifti: Are you sure it's not just silently dropping them? [23:40:05] if the regex doesn't apply it would drop them in my current configuration [23:40:35] gifti: So how do you know there were no misses? [23:41:33] i compared the script logs to the chat window [23:41:57] but of course i could also code filtering misses [23:45:04] a930913: you didn't translate that properly [23:45:18] :o [23:46:10] when you match for literal \x02 being there 0 or more times, you have to group it into a subexpression [23:47:29] gifti: Is there a reason for the * not being ? btw? [23:47:56] that seems to be a difference between tcl and other implementations [23:48:17] or not [23:48:30] i didn't know that that even existed [23:48:44] gifti: Looks better, but is only partially matching it. [23:49:50] gifti: It's only matching up to the user. [23:50:30] o.o [23:51:42] do you have another screenshot? [23:52:39] gifti: One sec. [23:53:55] gifti: http://i.imgur.com/9HLTVjw.png [23:55:10] a930913: try substitute .*? with .* please? [23:55:15] gifti: Ah, removed the.. Yeah. [23:56:23] hm, seems weird to me that this matched the other lines fully