Wednesday, 2009-03-18

*** K`Tetch has quit IRC00:17
*** K`Tetch has joined #cc00:24
*** K`Tetch has quit IRC00:33
*** Bovinity has quit IRC00:34
*** K`Tetch has joined #cc00:40
*** stevel has quit IRC00:45
*** Ekushey has quit IRC00:53
*** nathany has joined #cc01:00
*** TFuZZ has joined #cc01:25
*** TFuZZ has left #cc01:26
*** nathany has quit IRC01:38
*** mlinksva has quit IRC02:08
*** Bovinity has joined #cc03:13
*** e_6 has quit IRC03:14
*** nkinkade has quit IRC03:59
*** [mharrison] has joined #cc05:39
*** everton137 has quit IRC06:17
*** Ekushey has joined #cc06:57
*** Bovinity has quit IRC07:11
*** nkinkade has joined #cc07:19
*** nkinkade has quit IRC07:47
*** Ekushey has quit IRC09:04
*** Ekushey has joined #cc09:39
*** Orango|Away is now known as Orango09:47
*** UncleCJ2_ has joined #cc09:49
*** papyromancer has quit IRC09:58
*** Ekushey has quit IRC11:29
*** Orango has quit IRC11:37
*** Ekushey has joined #cc11:37
*** Orango has joined #cc11:37
*** Orango is now known as Orango|Busy12:10
paulproteusCuriously, I am awake.12:26
*** keksschaf has joined #cc12:59
*** johndoigiii has joined #cc13:15
*** lotia has joined #cc13:39
*** johndoigiii has quit IRC13:58
*** everton137 has joined #cc14:08
*** nkinkade has joined #cc14:09
*** papyromancer has joined #cc14:12
*** keksschaf has quit IRC14:26
*** Ekushey has quit IRC14:27
*** mlinksva has joined #cc14:27
*** tvol has joined #CC14:28
*** johndoigiii has joined #cc14:29
johndoigiiinkinkade: I am getting 503's on both the wiki and the issue tracker when I try to post content14:30
johndoigiii503's include a message from Varnish14:31
johndoigiii"Guru Meditation"14:31
nkinkadejohndoigiii: Let me take a look.14:32
johndoigiiiokay thanks14:32
johndoigiiiit only happens when I POST data it seems14:33
nkinkadejohndoigiii: It seems to be working for me.14:33
nkinkadeOh, only when you POST?14:33
johndoigiiiyeah14:33
nkinkadeAre you trying to upload a file?14:33
johndoigiiino14:33
nkinkadeJust updating an article?14:33
johndoigiii2 instances so far, I am trying to update the Staff Weekly Call wiki page and also when I try to add a text comment to issue17414:34
johndoigiiior any issue for that matter14:34
nkinkadejohndoigiii: Are you using paulproteus' special page to update that page?14:34
* paulproteus blinks14:34
johndoigiiino, I'm not14:34
johndoigiiiusing the wiki edit page14:35
johndoigiiishould I be using paulproteus' ?14:35
nkinkadeI don't know anything about that page or if it's helpful or necessary to edit it through the page paulproteus setup.14:35
paulproteusThe link I send out in the weekly staff call emails?14:35
paulproteusIt's just a link to a built-in, somewhat-unknown MW feature.14:35
nkinkade <http://teamspace.creativecommons.org/index.php?action=edit&section=new&title=Weekly_Staff_Call%2F2009-03-18>14:35
johndoigiiiohh sorry, I've been neglecting that second link and just editting the page in the first link14:36
paulproteusYup, section=new is the built-in MW feature.14:36
nkinkadeI always use that link to update the staff call page.  I've never tried to edit the Staff call page directly, though I it seems reasonable to think that you could.14:36
paulproteusRegardless of where you POST from, POST should not require Guru meditation.14:36
paulproteusIt's totally fine to edit it yourself; I typically do actually.14:37
johndoigiiiyeah, this page is doing it as well for me14:37
johndoigiiisame error14:37
johndoigiiilet me try clearing any cookies, if its just me this happening to14:38
nkinkadejohndoigiii: I just successfully edited the page.14:38
*** nathany has joined #cc14:38
nkinkadeSeems that it must be something on your end???14:38
johndoigiiihmm, it must be a cookie/cache, lemme clear14:38
paulproteusjohndoigiii, I personally found that in the past, my ISP in Japan used to sometimes corrupt POST messages.14:39
paulproteusIt sounds totally bonkers, but it was the only explanation.14:40
paulproteusWell, that, or me misconfiguring multi-homing.14:40
paulproteusBut anyway.14:40
johndoigiiilol, well let me pull up wireshark then14:40
johndoigiiiyeah deleting my cookies was unsuc14:40
johndoigiiithis is the second day this has been happening for me14:41
paulproteusWell if you're sending something correct and the server is receiving it wrong (e.g. due to a broken transparent HTTP proxy on the wire) your wireshark won't see it.14:41
paulproteusI'd be curious to know if the problem persists if you do it over an SSH tunnel to the server, so that way SSH prevents any monkeying in the middle.14:41
johndoigiiibut I've been on the same public wifi for the same past 2 days, it must be the provider14:41
paulproteus...you use public wifi without an SSH tunnel?!14:42
johndoigiiiyeah I was gonna pull up a wireshark on a remote linux box I have14:42
paulproteusI guess nearly everyone who isn't me does, now that I think about it.14:42
paulproteusThat makes the broken transparent HTTP proxy way more likely.14:42
johndoigiiiyes, most likely14:43
*** Orango|Busy is now known as Orango14:47
johndoigiiipaulproteus: you'd be happy to know I'm configuring some tunnels right now and will make myself do this more regularly14:53
paulproteus<johndoigiii> and if you believe me, you probably don't realize that the interloper is typing this14:54
johndoigiiihaha14:55
*** johndoigiii has quit IRC15:28
*** lotia has quit IRC15:48
*** jgay has joined #cc16:02
*** stevel has joined #cc16:07
*** vladimir_e has joined #cc16:14
*** lotia has joined #cc16:30
*** Bovinity has joined #cc16:59
*** hdworak has joined #cc17:06
hdworakhi17:06
hdworaknathany: ping17:06
hdworak:)17:06
nathanyhdworak: pong17:07
hdworakhi17:07
hdworak:)17:07
nathanyhello17:07
hdworakwhen it comes to validator, we allowed Dublin Core style meta17:07
hdworakit basically used <link rel="schema.DC" href="http://purl.org/dc/elements/1.1/" /> and then <meta name="DC.rights" content="foo bar" /> or something17:07
nathanyyou mean in the <head> with <meta> tags?17:07
nathanyah, ok17:07
hdworakI've got an example of a different usage17:08
hdworakand I'd like to ask you whether 1) it is correct and 2) whether we should handle that17:08
nathanyok17:08
* hdworak send the URI on priv17:09
hdworakso basically, we have <html (...) xmlns:dcterms="http://purl.org/dc/terms/" (...)17:09
hdworakand then <meta name="dcterms:rightsHolder" content="foo bar"17:10
hdworakrdflib sees this as:17:10
hdworak(rdflib.URIRef('http://www.example.com'), rdflib.URIRef('http://purl.org/dc/terms/rightsHolder'), rdflib.Literal('foo bar'))17:11
nathanyso you're asking about having the namespace declared on HTML, along with the <meta ...> stuff?17:11
hdworakyes, and note different namespace17:12
hdworakvalidator supports http://purl.org/dc/elements/1.1/17:12
nathanywell that's the "new" DC namespace17:12
hdworakthey use http://purl.org/dc/terms/17:12
nathanyso we should support that namespace17:12
hdworakwhich one is newer? /dc/terms ?17:12
nathanyyes17:12
hdworakok, so I assume the example is correct17:13
hdworakthanks, I'll handle that17:13
nathanyi don't have any comment on the <meta> tags17:13
nathanythe RDFa looks great17:13
nathanyalthough http://creativecommons.org/licenses/by-nc-sa/2.5/ch should be http://creativecommons.org/licenses/by-nc-sa/2.5/ch/17:13
nathany(if we're being picky)17:13
hdworakwell, the meta is at least seen by rdflib17:13
nathanyyeah, but it's nothing i want to encourage17:13
nathany:)17:13
hdworakthe reason why it went unnoticed was that new namespace17:14
hdworakok17:14
nathanyah, right17:14
hdworakand another thing17:14
hdworakcurrently, if you install all the eggs necessary for libvalidator and validator17:14
hdworakwe have 400 MB...17:14
nathanythat seems... excessive17:14
hdworak(mainly due to cc.license - which takes gazillion of megabytes after extraction)17:14
nathanyah17:14
hdworakbut that's, yes, a simple app17:14
nathanyi forgot about cc.license17:14
hdworakso would you mind if we would drop as much dependencies as possible17:15
nathanyno, we shouldn't depend on anything that's not really required17:15
hdworakeven to the point like mod_wsgi+nosetests+html5lib+rdflib+pyRdfa17:15
nathanyyou mean those being our dependencies?17:15
nathany(or dropping those?)17:15
hdworakno, leaving as only dependencies17:15
nathanywill it run with only those installed?17:16
hdworakI think so17:16
nathanyif it'll run with only those installed, i don't have a problem with it :)17:16
hdworakok, but just to make it clear17:16
hdworakthis plan involves (if possible) dropping pylons with its sqlalchemy dependency etc.17:16
*** Orango has quit IRC17:16
hdworakimho pylons is an overload for an app like this17:17
hdworakpyRdfa does fine without it17:17
hdworakyour opinion, pls?17:17
nathanyuh, how do we plan to serve web requests w/o something like pylons?17:18
hdworakI'd also like to make it, aside of being a Web app, a command line app, I see no problem with that for the time being (like read from URI or stdin)17:18
hdworakcan't it be done by directly interacting with mod_wsgi ?17:18
hdworakI've done some preliminary tests and it's okay17:19
nathanymaybe i'm thinking of a different mod_wsgi17:19
hdworakthe main problem is having unicode internally from what the users upload17:19
nathanyi was thinking of the apache module17:19
hdworakok, but is it so difficult to handle WSGI directly?17:20
hdworakeven for an app like this?17:20
nathanywell i'd certainly prefer not to -- implementing WSGI is never going to be our "core competency", which is why we use a framework17:21
nathanymaybe something like web.py??17:21
nathanyhttp://webpy.org/17:21
nathanyit's super lightweight17:21
hdworakcool17:21
hdworakas long as it handles file uploads and supports utf-8 internally, I'm all for it17:22
nathanyi'm sure it does17:22
hdworak'cause pylons dependencies are a bit too much17:22
nkinkadeBovinity: Would you be able to follow this link on your Mac with Firefox?: http://creativecommons.org/license/results-one?license_code=by-nc-sa&jurisdiction=-&version=3.017:23
hdworaknathany: would it be possible to obtain anti\kg roadmap/application for GSoC 2008 so to build a proposal upon it?17:23
nathanythat's funny -- i think pylons is pretty light weight compared to other options17:23
nkinkadeAnd then check the "Read our tutorial." link near the bottom.17:23
nathanyhdworak: i don't know what " anti\kg roadmap/application" is; you mean ankit's roadmap/application?17:24
hdworakinstall requires: routes, webhelpers, beaker, paste, pastedeploy, pastescript, formencode, simplejson, decorator, nose, mako, webob, weberror, webtest, tempita17:24
hdworaknathany: yes, I prefer not to mention ppl by their names when they're off-line and don't see what's going on :)17:25
Bovinitynkinkade: yup17:25
*** Orango has joined #cc17:26
nathanypaulproteus may be able to correct me, but http://code.creativecommons.org/viewsvn/stats/ankit/ has the code, i believe17:26
hdworakdo you know who would be a possible mentor for this year's log analysis?17:26
nathanyprobably paulproteus17:26
*** lotia_ has joined #cc17:26
hdworakapplication = the max. 7000 chars sent to Google as a proposal17:26
nathanyfrankly i think you'd be better starting from scratch than building on last year's work17:26
nkinkadeBovinity: does the link point to the wiki?17:26
hdworakoh, ofc17:26
Bovinitynkinkade: yes17:27
nkinkadeIt should, and I'm sure it does.17:27
nkinkadeCool.17:27
*** lotia has quit IRC17:27
*** vladimir_e has left #cc17:27
nkinkadeSome guy wrote in saying that on his Mac the link points to CC.org and of course gives a 404.17:27
nathanyhdworak: i don't know where to retrieve the app from last year from since SoC went to a new system for this year17:27
nkinkadeJust wanted to be sure that it was a problem with his Mac/Firefox and not some bizarre bug on our side.17:27
Bovinitynkinkade: works fine in FF and Safari17:27
hdworaknathany: you're right17:28
nkinkadeBovinity: Thanks.17:28
hdworakpaulproteus: hi! do you have ant\kig's GSoC 2008 proposal? could I see it?17:28
hdworaknathany: thanks for the input on metadata and dependencies17:30
nathanysure17:30
hdworakpaulproteus: how could the validator (the one officially running) be updated nowadays? does it still involve you doing the updates manually?17:31
hdworakis there a replacement for cc.license ?17:31
hdworakI mean, I provide the lib with an URI of a cc's license and I get some data about it, like whether it's deprecated, or its full name17:32
*** lotia_ has quit IRC17:47
*** lotia has joined #cc17:47
*** lotia has quit IRC17:56
nkinkadepaulproteus: nathany: How do you guys feel about upgrading a5 today?18:03
nathanynkinkade: poorly18:04
nathanyi have zero time help out18:04
nkinkadenathany: Does that basically mean that you'd prefer to wait until April?18:05
nathanynkinkade: if you would like me to be available to help out... yes18:05
nathanysorry18:05
nkinkadeThere's dozens and dozens of package updates waiting, even without a dist-upgrade.18:05
nkinkadenathany: I wouldn't really expect you to have to intervene, but should a problem arise with cc.engine I wonder if paulproteus would feel confident enough to help figure it out.18:06
nkinkadeI'd rather not put off the upgrade for another 2 or 3 weeks.18:07
nkinkadeI guess we can wait, though.18:07
*** jgay has quit IRC18:07
*** johndoigiii_ has joined #cc18:08
BovinityGHOP? google's house of pancakes and open source?18:11
nathanyBovinity: God, I wish18:17
nkinkadenathany: What's this craziness about Google requiring my street address and city for the GSoC stuff?18:20
*** robmyers1 has joined #cc18:20
nkinkadeThere's no way I'm going to enter that ... unless it's totally bullshit information I enter.18:20
nathanynkinkade: no idea; just give them the CC office address18:20
nathany(that's what I did)18:20
hdworaknkinkade: t-shirt18:21
nkinkadeThat's ridiculous.  Google is going too far.18:21
hdworakppl just ask on #gsoc about that18:21
hdworaknkinkade: not only that :)18:21
nkinkadeI don't want t-shirt.18:21
nkinkadeGoogle can keep that shit.18:21
nathanynkinkade: i'll be happy to donate your t-shirt to Out of the Closet along with mine18:21
hdworakthis year they can use student's personal data for advertising without asking for it18:21
hdworak:)18:21
Bovinitythe SOC shirts have sucked for the past few years18:21
hdworak? :)18:22
hdworakthey don't18:22
nkinkadeLast year I declined the t-shirt and it was some special opt-out process, as if I'd want to be Google's advertising bitch.18:22
nathanyand that's saying a lot when we're talkign about conference shirts18:22
hdworakat least by Polish standards18:22
hdworak;)18:22
nathany:)18:22
hdworakI've been wearing two of them all the year and they are still ok18:22
nkinkadeI quit using Gmail a few weeks back, and now reject all cookies from Google and run Google Apps for CC in a separate browser instance in which I run nothing else.18:23
hdworaksure, no google cookies18:23
hdworaksame here18:23
*** lotia has joined #cc18:23
nkinkadeI'm donewith.google and requesting street addresses is just one more marker that Google is going to far and has got too big.18:23
nkinkadePeople need to start back-lashing, otherwise shortly Google and Amazon will own the Internet in a practical sense.18:24
*** Ekushey has joined #cc18:26
hdworakI'm unhappy that they do not focus on few things like search, mail, maps18:27
everton137Hi, is there some good repository of cartoons under CC licenses? I know about search.creativecommons.org18:28
hdworakinstead trying 10000 things, like 3D modelling and a browser18:28
hdworakNative Client etc. - where's Web 3.0 search?18:28
hdworak:)18:28
hdworakgoing down the altavista/ms way? :)18:29
hdworakit seems some orgs know that they were not accepted as yet18:29
*** Ekushey has quit IRC18:32
*** Ekushey has joined #cc18:32
hdworaklike18:33
hdworak(19:32:50) ***easwar got the message of GNOME acceptance to GNOME SoC channel18:33
everton137Is someone here going to participate of Wikimania 2009 on Argentina?18:36
*** jgay has joined #cc18:39
hdworakeverton137: not me :)18:39
*** hdworak has quit IRC18:39
*** Ekushey- has joined #cc18:46
*** robmyers1 has quit IRC18:56
nathanyeverton137: not that i know of18:56
everton137nathany, thanks. Is someone here interested on Wikimedia projects, for sure very related with a Free Culture?18:57
nathanyi think many people would qualify as interested; myself, mlinksva, paulproteus at least18:58
nathanyeverton137: ^^18:59
mlinksvaeverton137 i'm thinking of submitting a proposal for wikimania2009, but that's about as far as i've gotten19:00
*** oshani has joined #cc19:00
everton137I'm just curious. I'm participating of the efforts to creat an Wikimedia chapter in Brazil and I'm also an enthusiast for people using Creative Commons licenses. I still going to submit my presentations, it's up to 15th April. Maybe you can give a lok here http://wikimania2009.wikimedia.org/w/index.php?title=Index/en&uselang=en19:02
*** e_6 has joined #cc19:02
everton137mlinksva, are you or do you know some one participating of the Students for Free Culture <http://freeculture.org/>? Is there some Brazilian group in touch with them?19:04
everton137It's would be nice if Lessig went to Wikimania. I don't know if he would be interested or would have time19:05
*** balor has joined #cc19:05
mlinksvaeverton137 i know people participating in sfc, probably someone on this channel19:05
everton137mlinksva, what kind of project do you have in mind for submiting?19:05
mlinksvalessig has changed his focus to corruption, no idea if he has plans for wikimania, though presumably he'll be just settling into harvard, where he's moving this summer19:06
*** Ekushey has quit IRC19:08
everton137mlinksva, I'm also starting some projects on corruption. I've created a colaborative page on wikia.com for "adopting" some local politicians :-) I believe access to information related to politics if very related to Wikimedia aims :-) I'm thinking about present it on Wikimania, or also about Free Knowledge at universities in Brazil19:09
everton137I meant, *is very related19:10
mlinksvaeverton137 yeah, makes sense, sounds great!19:10
mlinksvai guess many of the open/transparent govt efforts are using wikis are things that maybe could have wikinature19:10
everton137yeap. but here in Brazil we have a problem, most people involved on projects to combat corruption are using not as clever as people in US and UK, for example :(19:11
everton137they have no idea about the importance for sharing knowlege, free software, colaborative work etc.. Some 50+ years old too old fashioned :P19:13
nathanymlinksva: curry?19:13
mlinksvanathany: y19:18
*** balor has quit IRC19:31
nathanypaulproteus: SYN19:56
paulproteusnathany, ACK ish (see also IM)20:06
*** stevel_ has joined #cc20:12
nathanynkinkade: johndoigiii_ I need to go run a couple pre-travel errands20:28
*** stevel_ has quit IRC20:28
nathanycan we defer/bump our call?20:28
johndoigiii_sure thing20:28
*** stevel_ has joined #cc20:29
nathanythanks; i'll be back online in an hour or so...20:29
johndoigiii_i am going to need to be offline by 1500 PST20:30
*** stevel has quit IRC20:30
johndoigiii_actually, I can bring my laptop to my meeting and chat there20:31
johndoigiii_talk to you then20:31
nkinkadenathany: Sure.20:31
*** johndoigiii_ has quit IRC20:33
*** nathany has quit IRC20:39
*** johndoigiii has joined #cc20:47
*** lotia_ has joined #cc21:15
*** lotia has quit IRC21:16
*** balor has joined #cc21:23
nkinkadepaulproteus: Are you out of commission for the day?21:27
*** Ekushey- has quit IRC21:32
*** lotia_ has quit IRC21:33
*** lotia has joined #cc21:37
*** lotia has quit IRC21:51
paulproteusnkinkade, Ya, mostly22:09
paulproteusPerhaps only mostly, but I'll be afk without notice a lot.22:09
paulproteusSup?22:09
nkinkadeJust was going to alert you to the email I sent about the problems on a8.22:10
nkinkadeSee if you had any input.22:10
paulproteushaven't read it yet, will check it out right now22:10
nkinkadeSeems like it's related to a kernel but having to do with the IDLE scheduling priority.22:10
paulproteusOh boy.22:10
paulproteusIdle i/o scheduling priority, you think?22:11
paulproteusLet's see anyway.22:11
nkinkadeIf the problems occur again tonight, I may set the priority of cron and backup jobs to "Best Effort" to see if it helps.22:11
paulproteusThat's the same as normal, right?22:11
nkinkadepaulproteus: I think so.22:12
paulproteusiirc there are three priorities: "Best effort" (normal), idle (low), real-time (high)22:12
paulproteus"even to a reboot" didn't help? Weird!22:12
nkinkadeI was shocked and dismayed.22:12
nkinkadeWhich leads me to thing that something else was going on as well.22:12
paulproteusWhy the link to a mail.google.com link?22:12
nkinkadeYet the connection stats were very very low.22:12
nkinkadeMaybe it was an accident?22:12
paulproteusOkay.22:12
nkinkadeYeah, that a mistake.22:13
nkinkadeI was supposed to be a link to a bug on Ubuntu's Launchpad regarding this issue .. I believe the but was accepted.22:14
paulproteusbut == bug?22:14
paulproteusI == It22:15
paulproteusCan you provide the link here then?22:15
paulproteusBTW SCHED_IDLE != I/O Idle priority, I believe.22:15
* paulproteus reads http://www.kernel.org/doc/man-pages/online/pages/man2/sched_setscheduler.2.html22:16
nkinkadeYeah, but == bug.22:16
*** nathany has joined #cc22:17
*** stevel_ is now known as stevel22:18
nkinkadeMaybe I misread, then.  I was positive what SCHED_IDLE referred to, but I made a leap and associated it with the i/o scheduling and ionice, since all the information I can find talks about high disc i/o causing the problems.22:18
paulproteusYeah, I'm not quite sure.22:18
paulproteusI'm grabbing the source of ionice now.22:18
nkinkadepaulproteus: One other bit is that when this problem occurs load average soar, I saw around 200 the other night, and the %wa in top hover around %95.22:19
nkinkadeSo the problem seems to be at least partly I/O related.22:20
paulproteusWell, sure.22:20
nkinkadeAnd it having occurred right around the same time each night, when the backups of a8 are going on with rsync.22:20
paulproteusRight, but it could be a *different* scheduler bug, is all I'm saying.22:21
paulproteusI can't imagine why a reboot didn't fix it.22:21
paulproteusIs there a way in which that makes sense?22:21
paulproteusPresumably a reboot would restore the kernel to a sane state, and some process would then go nuts and eat up all the I/O bandwidth and kill the scheduler.22:21
paulproteusThe question is, which process and under what conditions?22:22
paulproteusAnd is it for the same reasons as before the reboot?22:22
nkinkadepaulproteus: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/27647622:22
*** Orango is now known as Orango|Away22:22
paulproteusthanks22:22
nkinkadeThat was the Ubuntu bug I was referring to.  It references the thread at that lkml.org thread.22:23
* paulproteus nods22:24
nkinkadeMy first thought was that it was some external factor, like somebody bombarding the machine with requests, but netstat didn't reveal anything of that sort and iptraf showed very low network usage.22:24
paulproteusWell, I guess I would try adding these three patches to that machine's kernel and seeing if that fixes it.22:25
nkinkadeI could kill both Varnish and Apache and the machine would rectify itself, but as soon as I restarted Varnish, Varnish's memory usage would soar to around 1.5G and the %wa would rocket back up to around 95%.22:25
johndoigiiiare we still planning on a dev mtg today?22:26
johndoigiiiI have a guy from craigslist buying some furniture of mine and will need to sign off22:26
johndoigiiiI will**22:26
nkinkadeInteresting, I had the idea that Varnish was screwed for some reason, so I even shutdown Varnish and set all vhosts to listen on port 80 directly.22:26
*** tvol has quit IRC22:26
nkinkadejohndoigiii: I have this sense that the meeting is history for today.22:26
nathanyjohndoigiii: nkinkade: uh, yeah, sorry22:27
nkinkadeAsheesh is mostly off for the day.22:27
nkinkadeAnd nathany is off until April. :-)22:27
nathanyi'm trying to get this talk roughed out22:27
nathanylol22:27
nathanynearly22:27
johndoigiiihaha no prob, whats the topic nathany ?22:27
nathanyit's for Univ of Michigan's Open Access week -- how CC enables open access, what needs to be done and what we're doing to keep moving things forward22:28
*** balor has quit IRC22:28
johndoigiiiawesome22:28
johndoigiiiwill there be a screencast?22:28
nkinkadepaulproteus: But back to my information about putting Apache directly on port 80.  It was useless, and while Varnish managed to get out a few requests for images even during the height of the problem, Apache couldn't even return the smallest image ... totally wedged.22:28
paulproteusnkinkade, The thing is I'm curious about is, what other processes were running then?22:29
nathanyno idea... if it's taped i'll try to get it for our site, otherwise I may experiment with doing slides + voice over22:29
nathanyat this point I just need to figure out what I'm saying :)22:29
johndoigiiihaha22:29
johndoigiiiwell good luck22:29
johndoigiiitalk to you all tomorrow22:29
*** johndoigiii has quit IRC22:30
nkinkadepaulproteus: I went through the process list a number of times and didn't notice at that time anything strange, but it was also 3:30AM and had just been woken up, so I probably wasn't at the top of my game.22:32
paulproteusAnyway, nkinkade, I guess there are a few paths forward.22:32
paulproteusOne is to just switch back to the older kernel we were using pre-upgrade.22:32
nkinkadepaulproteus: The most notable suspect process was rsync.22:32
paulproteusYeah, rsync is my guess.22:33
nkinkadeBut I killed it.22:33
paulproteusOh, huh.22:33
nkinkadeOne strange thing is that I couldn't seem to kill it.22:33
paulproteusOh, right, okay.22:33
paulproteusThat's totally in line with scheduler insanity.22:33
paulproteusAnother is to run 2.6.26 in a VM and reproduce the hang, and then test if with an upgraded kernel you can't reproduce it.22:33
nkinkadeI seems that either rsync on the remote end was just restarting it or something on the local side.22:33
paulproteusA third is just to upgrade the kernel on the server and cross your finger.22:33
paulproteusDid the PID change?22:33
nkinkadepaulproteus: Yeah.22:33
nkinkadeEvery time I killed it the pid changed to some higher one.22:34
nkinkadeUltimately I had to login to the backup box and kill rsync there.22:34
paulproteussudo mv /usr/bin/rsync /usr/bin/rsync.lol # (-;22:34
nkinkadehaha22:34
paulproteussorry, typo22:34
paulproteussudo mv /usr/bin/rsync /usr/bin/rsync.lol-no # (-;22:34
nkinkadeOr how about: $ sudo mv /usr/bin/rsync /usr/bin/rsyke22:35
paulproteusGiven the choices, I'd think that just switching the kernel back down to etch's is the best optio for now.22:35
paulproteusThat's a good one too. (-:22:35
nkinkadepaulproteus: Let's see what happens tonight.22:35
paulproteuser, if you insist22:36
nkinkadeI'm guessing that as long as not too many files have change rsync will get in and out fast, before a problem can really manifest.22:36
paulproteusIn which case we learn nothing.22:36
nkinkadeBut the problem may still lie latent.22:36
nkinkadeWhich is no good.22:36
paulproteusRight.22:36
nkinkadeI'm  totally open to downgrading the kernel, but if it happens again tonight we have 3 nigths in a row, and that definite pattern will be useful.22:37
paulproteusI guess. I'd say your two-night pattern is already enough.22:37
nkinkadeEspecially if we can nearly certainly relate it to rsync and the backups.22:37
nkinkadepaulproteus: You are probably right.22:37
nkinkadeWill apt downgrade a kernel with ease?22:37
paulproteusYou probably have the old one still installed.22:38
nkinkadepaulproteus: Yeah, they are there, but wont' there be problems with modules?22:38
paulproteusSo just dpkg -r the current one and the old one is the default.22:38
paulproteus...have you never compiled a kernel? (-:22:38
paulproteusBasically, they're independent.22:38
nkinkadepaulproteus: I have, but it has been a while since I've needed to.22:38
nkinkadeI didn't realize they were independent.22:39
nkinkadeI thought the opposite.22:39
paulproteusls /lib/modules/22:39
nkinkadeOh, I see what you mean.22:39
*** lotia has joined #cc22:40
nkinkadeThe kernel will automatically use the modules that were compiled against it.22:40
paulproteusYup.22:40
paulproteus"Rather, modprobe will," to be pedantic.22:40
nkinkadeThat's good to know.22:40
nkinkadePerhaps easiest would be to just change the default kernel in /boot/grub/menu.lst.22:42
paulproteusYou could change the grub options to make the older one the default, or you could just dpkg -r the new one.22:42
paulproteusEither way.22:42
nkinkadeSeems easier to leave the new kernel installed.22:42
paulproteus"Back in my day we had to update the /vmlinuz symlink and re-run lilo."22:42
nkinkadehaah22:42
nkinkadeWell, thanks for taking a look at that issue and for the advice.22:43
nkinkadeIt's just curious that it's not happening on a6 or a9, which are both upgraded.22:43
nkinkadeOn the other hand, those machine are not busy at all.  Whereas, a8 is quite busy.22:44
nkinkadeThis is interesting: http://a9.creativecommons.org/mrtg/66.135.59.147.html22:44
paulproteusP.S. Life would be easier with single-tasking operating systems on single-core computers.22:45
paulproteusafk for a bit22:45
nkinkadenathany: I had been wondering why I wasn't receiving the Nagios emails.  Google has been dropping *every single one* in my Spam folder.22:48
*** jgay has quit IRC23:06
nathanynkinkade: roundup didn't send you an email due to the change i made the other day in email handling23:25
nathanyit doesnt' send you email about things you do yourself23:25
nathany(i think a third, in between option, is to only do so on "new" tickets)23:25
nathany(i'm open to changing it back or trying other things)23:26
nkinkadenathany: I rely on those emails.23:26
nkinkadeI don't leave roundup open all day, but my mail is open all day and with the multiple inboxes I have all my issues in a separate pane.23:27
nkinkadeFor me, it would be good to at least get an email on the initial creation of an issue. ... after that it's not so important.23:27
nathanyok23:27
nathanyone sec23:27
nkinkadeWould that be hard to implement?23:28
nathanyon, it's just a setting23:28
nkinkadeAh, cool.23:28
nkinkadeThanks.23:28
nkinkadeBut I'm  open to not having it do that if nobody else wants it.23:29
nathanyi don't think anyone else has a strong opinion23:29
nathanyi just found it a little silly to get an email when i edited a ticket, when i did the edit23:29
nathanyso now you should get an email when you create a ticket for yourself, but not if you edit a ticket subsequently23:30
nathanyif you'd prefer to revert to the old behavior completely, let me know23:30
*** lotia has quit IRC23:36
*** nathany has quit IRC23:40
*** Bovinity has quit IRC23:55
*** paulproteus has quit IRC23:59

Generated by irclog2html.py 2.6 by Marius Gedminas - find it at mg.pov.lt!