13:30:02 #startmeeting The release 13:30:02 Meeting started Tue May 25 13:30:02 2010 UTC. The chair is mmcgrath. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:30:02 Useful Commands: #action #agreed #halp #info #idea #link #topic. 13:30:20 #action add task to hide -Alpha and -Beta releases from the mirrorlist on release day 13:30:22 I'm going to push spins.fp.o ahead of time 13:30:40 #actiln getting more seeders ready 13:30:40 but I don't want to do rewriterules 13:30:46 nirik: how's your torrent stuff looking? 13:30:48 smooge: Redirect 301 /originalpath /newpath 13:30:53 * mdomsch has a seeder running 13:30:58 oh 13:31:01 No mod_rewrite needed :-) 13:38:48 #info Announcement for mailing list is queued, checked all links 13:42:07 Heh, a syncStatic takes far longer in production than in staging 13:42:37 morning 13:43:08 Hiya Oxf13 13:43:09 oh, let me get some ini files in place for torrents 13:43:28 * stickster has PR folks on hold for the Red Hat press release 13:43:36 Maybe it has to do something with load average: 37.93, 35.87, 32.98 :-) 13:43:46 ricky: where you seeing that load average at? 13:43:48 wowsers 13:43:53 bapp01 13:44:08 (The one place where it's normal, I guess) 13:44:28 eh, that's still a bit high, it's all cpu time too 13:44:47 Maybe time for a bapp02 at some pint 13:44:48 **point 13:45:17 There's crawling and transifex pulling going on 13:45:35 Website build took almost 13 minutes there, takes less than 5 on app01.stg 13:45:47 ricky, mmcgrath look at http://fpaste.org/L7qM/ and let me know what I need to do 13:46:07 smooge: Did you add that file in fedora-docs/manifests/init.pp? 13:46:15 not yet 13:46:23 ok, torrent inis in place 13:46:27 (Just copy the fedora-docs.conf block from above and replace it with redirects.conf) 13:47:01 get that redirects pushed out asap, we don't want that syncing seconds after the announcement goes out. 13:47:35 we can kill crawlers if we need 13:47:36 * stickster can hold announcement until things get confirmed 13:47:48 ricky, thanks for catching that 13:47:50 tx looks like the killer right now 13:47:55 Should I give PR folks a target time after 10:00, like 10:15 or 10:30? 13:48:05 mdomsch: Nah, not a problem, the websites build is finished now 13:48:31 smooge: I sent my +1 13:48:41 smooge: I imagine mmcgrath is happy to +1 too 13:48:50 +1 13:49:12 ricky: Oxf13: we on pace for a 10:00 release? 13:49:20 I am 13:49:21 Websites are ready to go 13:49:46 ricky, pushed.. needs a puppetd -t on the proxy servers 13:50:11 * smooge wonders if a func call would work that way 13:50:17 smooge: I just ran this: 13:50:18 make HOSTS="proxy1 proxy3 proxy4 proxy5 proxy6 proxy7" push 13:50:21 Oxf13: ricky: mmcgrath: smooge: OK, shall I wave the checkered flag for the PR folks to push promo stuff to redhat.com and the press release for a 10am go? 13:50:24 Handy to know for future reference 13:50:28 In puppet/ 13:50:50 interesting 13:50:51 stickster: as far as I'm concerned, yes. Most of my stuff was done last night. 13:50:58 how are we looking on amount of mirrors open? 13:50:59 make push HOSTS :) 13:51:10 smooge: It works via snmp, might not work for some machines due to firewall rules, but it's very handy for doing all the proxies 13:51:18 Oxf13: I'll re check, it was 90% earlier 13:51:39 Hm, proxy7 isn't in func 13:51:44 But it isn't live either, so no problem 13:52:03 Er, I take that back 13:52:03 Adding it now 13:52:10 hmmm I thought I got it in func 13:53:20 Aaaand I have no idea how to add it. 13:53:20 Oxf13: yeah still sitting around 90% 13:53:26 I can just run the commands manually 13:54:51 so I think we're just waiting for the ricky vote on pulling the trigger 13:55:00 Oh, nothing blocking from here 13:55:16 Just let me know when you want the website pushed 13:55:48 mmcgrath, who is that last 10%? 13:55:52 Oxf13, when was the bitflip? 13:56:01 we're starting to see an up-tick in proxy requests - http://mmcgrath.fedorapeople.org/monitoring/proxy-requests.html 13:56:18 mdomsch: http://fpaste.org/cN3t/] 13:56:21 oops - http://fpaste.org/cN3t/ 13:56:24 mdomsch: I did it around midnight pacific time 13:56:33 nearly 7 hours ago 13:56:45 ricky: lets push the website now 13:56:55 OK, running 13:57:19 And with that, I'm going to have Red Hat move their bits out too -- it will take 10-15 minutes for them to show up as well. 13:57:44 * stickster will save ML announcement until websites are ready 13:58:31 OK, website's out, doing spins now 13:58:42 * mmcgrath does a link check 13:59:42 ricky: FYI, http://stg.fedoraproject.org/en/get-fedora != http://fedoraproject.org/en/get-fedora 14:00:04 They look the same to me 14:00:04 ricky: nm, there it goes. 14:00:12 Ah, maybe you hit proxy6 14:00:21 I'm syncing manually there now as well, func didn't hit it either. 14:00:36 It hit it fine for spins though 14:00:48 Weird? 14:01:36 mmcgrath: ricky: OK to release ML announcement now? 14:01:49 * ricky is fine with it 14:02:38 stickster: looks good. I got 100% link coverage 14:02:53 * mmcgrath finds the clickstream of the new design a bit confusing 14:04:17 A result of trying to include everything :-( At least it's not on the main download page though 14:04:35 ricky: yeah, it's still much better. Don't get me wrong 14:04:51 https://admin.fedoraproject.org/collectd/bin/graph.cgi?hostname=proxy01;plugin=apache;type=apache_bytes;begin=-3600 <-- weird 14:05:02 Wow :-) 14:06:24 stickster: why "http://get.fedoraproject.org?F13an" instead of "http://get.fedoraproject.org"? 14:06:32 For link tracking purposes 14:06:36 So we know where people came from 14:06:41 lol 14:07:18 rsc: http://lists.fedoraproject.org/pipermail/marketing/2010-May/012918.html 14:07:19 rsc: feel free to use a tinyurl or bit.ly or soemthing 14:07:23 Date: Tue, 25 May 2010 10:03:40 -0400 14:07:26 Have we collected any data from thos eyet? 14:07:28 **yet 14:07:33 that might be the closest to 10:00 am release we've ever done (: 14:07:48 herlo: rsc: There are links prepared at https://fedoraproject.org/wiki/Link_tracking 14:08:04 Heh, running the website sync half an hour in advance helped :-) 14:08:04 herlo: rsc: Using 100% free software ur1.ca site, no less :-) 14:09:05 ricky: Not yet -- we need a more solid Marketing SOP for them to make sure we're measuring apples to apples 14:09:16 OK 14:09:22 stickster: yeah, figured :) 14:09:29 Anybody have a .announce prepared? 14:10:03 $ rm -rf 13.RC3/ 14:10:03 rm: cannot remove `13.RC3//Fedora/i386/iso/.nfs000000000094f46100000012': Device or resource busy 14:10:15 mmcgrath: more wackyness with that secondary1 storage 14:10:28 ricky: No, but I can paste my recent dent 14:11:01 Sounds good, go for it 14:11:02 ricky: Sound good? 14:11:07 ah, ok :-D 14:11:42 done 14:11:45 Oxf13: looking 14:13:09 Oxf13: that's interesting. I assume that's not somethign you actually created? 14:13:17 that is correct 14:13:30 marek: ping 14:13:50 Announcement from my owner (stickster): Happy Fedora 13 Release Day! http://ur1.ca/03bfg -- to download: http://ur1.ca/03ajw 14:15:05 mmcgrath: pong 14:15:08 mdomsch: just to confirm, redirects for releases/13/ are disabled correct? 14:16:00 Oxf13, they still point at development/13 but the content is the same 14:16:05 I can drop them now 14:16:10 marek: hey, just created a weird ticket. I'll give you the ID if you want to look (when I get it back) 14:16:10 Note that there are pretty much 0 torrent links at fp.o 14:16:17 please do, I just closed the ticket to make that happen 14:16:25 So there may be more via mirrors this time. 14:17:54 np 14:18:16 we're serving around 380 requests / second at the moment 14:20:36 I'm getting mirrors.txt 14:20:42 Er, releases.txt 14:21:31 ricky: thanks I was just about to do that :) 14:23:15 Oxf13, done 14:23:22 will take effect top of the hour 14:23:27 marek: - INC000000198176 14:24:18 Thanks to whoever got func on proxy7 14:25:20 mmcgrath: let me find someone to have a look at it 14:29:15 marek: thanks, appreciated. not a huge major deal or anything but pretty weird :_/ 14:29:26 :) 14:30:39 what can I help with at the moment 14:31:13 smooge: right now we're just trying to pay attention to everything :) 14:33:11 mdomsch: barf, the anaconda redirect as method didn't get fixed :( 14:33:25 mmcgrath, ? 14:33:37 which redirect? 14:33:39 BFO is still using - method=http://download.fedora.redhat.com/pub/fedora/linux/releases/13/Fedora/i386/os/ 14:33:49 because download.fedoraproject.org doesn't work 14:33:50 ouch 14:34:00 oh, anaconda doesn't handle the redirect 14:34:04 yeah 14:35:00 so is BFO getting used so much taht d.f.r.c can't handle it? 14:35:16 not yet 14:35:16 of so, yea!, and "oh" 14:35:26 and I suspect it we'll be ok. 14:35:54 if we need to get real fancy we can have it serve that as a server side script that changes method on download. 14:36:27 mmcgrath: torrents are humming along here. ;) 14:36:55 * mdomsch seeding at only 200KB/sec 14:37:09 cap is above 750 14:37:18 hrm, my rsync from warthog9 is failing... should I just manually bit-flip for now, or are there new things being pushed as well? 14:37:32 Jeff_S, manual please 14:37:38 mdomsch: ok, thanks 14:37:59 mdomsch, how can I get the metalink files for the isos from mm? and do they include the torrent information? 14:38:39 heffer, the metalink files for isos are per-ISO, not grouped, and do not inlcude torrent info 14:38:56 http://fedoraproject.org/get-fedora-all 14:39:04 Ticket notification - infrastructure: Ticket #2144 (bug closed): Template modification (FedoraVersion 14:39:07 Not linked from anywhere at the moment, but it has everything 14:39:25 heffer, http://mirrors.fedoraproject.org/metalink?path=pub/fedora/linux/releases/13/Fedora/x86_64/iso/Fedora-13-x86_64-DVD.iso 14:39:33 mdomsch, too bad. because using them with aria2 in http/ftp/torrent mode would be really nice. 14:39:47 mdomsch, thanks for the link 14:39:51 heffer, just a small matter of code I haven't had time to write 14:40:01 mdomsch, okay :) 14:44:11 ricky: ping 14:44:18 mmcgrath: pong 14:44:34 ricky: can you fix the F13 release.txt name. It says (Branch) :) 14:44:55 * mmcgrath is going through and double checking everything 14:45:08 Blah, I really should have slept last night 14:45:11 Good thing you checked 14:46:29 ricky: hehe, you were too excited for the release ;) 14:49:01 https://admin.fedoraproject.org/collectd/bin/index.cgi?hostname=proxy3&plugin=apache×pan=3600&action=show_selection&ok_button=OK 14:49:15 keepalives are taking up a lot of our scoreboard at the moment. 14:51:55 hmm, we're seeing a lot of docs errors 14:52:23 stickster: who's someone good to talk about wrt docs errors? 14:52:49 ricky: hey got a minute? 14:53:31 mmcgrath: Sure 14:53:44 ricky: ssh to proxy3 and tail -f *error.log 14:54:02 mmcgrath: Probably best is rudi, but he's in BNE so he may have finally dropped off for some shuteye 14:54:10 :-/ 14:54:10 I can try to help as much as I can 14:54:26 We can make sure sparks is here too, as Docs team lead. 14:54:27 stickster: I'm not sure what is including this - http://docs.fedoraproject.org/es-ES/Fedora/13/html/Release_Notes/Common_Content/css/default.css 14:54:34 but it's refering to css docs that don't exist 14:54:46 so browsers try to request them and they don't return 14:55:14 Hm, I'm guessing that needs to be fixed on the publican side 14:55:32 mmcgrath: Should we put empty docs in for those? ricky is right, they're generated by the Publican brand bits 14:55:50 heh, that's a fun one - [Tue May 25 14:55:32 2010] [error] [client 67.207.201.165] (36)File name too long: access to /get-fedora to download a LiveCD, regardless of what operating system youre running. This will give you a working version of Fedora, complete with common applications, all running off your CD drive - your hard drive wont be touched at all.And when youre ready, installation is just a click away.\nWant an even more enjoyable way to use F 14:56:03 Hehe 14:56:11 wow 14:56:14 Mispaste, happens to me all the time 14:56:20 * stickster too :-) 14:56:33 Good thing my URLs usually don't have rm -rfs in them :-) 14:57:14 mmcgrath: I can generate empty files for those CSS as a temporary fix and put them in our docs/web git repo 14:57:15 mmcgrath: so the 404 on the docs site is pushing them to the error/welcome page? which means the docs are basically inaccessible I assume 14:57:31 ke4qqq: beats me, you'll have to talk to the docs guys. 14:57:34 ke4qqq: I wouldn't think so, just generating lots of other 404s in the logs. 14:57:42 I just loaded a bunch of docs with no problem. 14:57:57 stickster: lang.css is the problem - 14:58:04 The error pages should redirect to / 14:58:09 So nothing should be inaccessible 14:58:28 Well, except for that langs.css file :-) 14:58:47 http://docs.fedoraproject.org/es-ES/Fedora/13/html/Release_Notes/index.html <-- seems to work OK 15:01:37 mmcgrath: Talking on #fedora-docs about this now 15:01:48 ricky: yeah, I can see in yslow, it's actually downloading and trying to run the main page stuff for all css docs that are missing 15:01:59 which isn't breaking the page, but is very inefficient 15:02:03 stickster: thanks 15:02:10 If we have a testing doc for staging sites, I recommend adding this to the test measurements 15:02:14 mmcgrath: anything I can help out with as far as release-day heckticness? 15:02:32 mmcgrath: Shall I whip up a rewriterule? 15:03:25 ricky: I'm not sure. 15:03:39 It would fix the extra useless downloading 15:03:45 stickster: this is something that could be tested before it gets to staging 15:03:46 ricky: empty file wouldn't be easier? (/me doesn't claim to know the answer, just asking) 15:03:47 Any downsides? 15:04:02 ke4qqq: It's under a bunch of paths :-/ 15:04:09 ricky: I say hold off for now and let the docs team see if they can fix it and push it out. 15:04:23 Publican is designed to have each language have its own files and be self-contained 15:04:26 OK 15:04:29 it doesn't seem to be causing anything horrible. 15:04:58 ricky: here's one you could fix - [Tue May 25 15:04:42 2010] [error] [client 59.172.157.158] File does not exist: /srv/web/spins.fedoraproject.org/favicon.ico 15:05:03 This is a good lesson to avoid last-minute stuff next time. Hopefully we've mitigated most of the issues 15:05:07 OK 15:05:16 [Tue May 25 15:04:47 2010] [error] [client 67.207.201.163] (36)File name too long: access to /get-fedora to download a LiveCD, regardless of what operating system youre running. This will give you a working version of Fedora, complete with common applications, all running off your CD drive - your hard drive wont be touched at all.And when youre ready, installation is just a click away.\nWant an even more enjoyable way to use Fedora, risk-free failed 15:05:22 that one just popped up again :) 15:05:43 hah 15:05:48 Haha, from a slightly different IP too 15:05:51 Same guy, I guess 15:06:10 mmcgrath: twice??? is that a bad link somewhere? 15:06:38 The IPs are 2 apart, for what it's worth 15:06:54 ah, it's a bot. 15:06:57 * ricky would hope our link checking would have caught that - did a lot of that last night 15:06:59 +http://www.chainn.com/mxbot.html 15:07:01 whatever that is 15:07:06 Hah, smart bot 15:07:37 interesting 15:07:45 mmcgrath: Anything I can help out with, or are we good? I have a few minutes, then I'll be gone for ~45 minutes, then I'll have quite a while if we need anything 15:08:00 CodeBlock: aside from more torrent seeders I think we're in good shape at the moment 15:08:16 mmcgrath: I'm already seeding, grabbed it last night, heh ;) 15:08:27 * nirik is pushing about 4800KB/sec for torrents now. 15:08:32 I'll bump up the throttle on my vps a bit for it 15:09:06 I'd seed at SHSU but we block torrenting in and out, so... :-/ 15:09:22 * rigeld2 stabs his security guy 15:09:36 yeah torrent is dangerous 15:09:36 Whoever designed the idea of /favicon.ico.. 15:09:41 mmcgrath: actually I'm only seeding -live, I'll grab some of the others I guess 15:09:51 ricky++ 15:13:23 mmcgrath: Which torrents really need some seeders atm, any particular ones? 15:14:01 It's hard to tell, you can look at http://torrent.fedoraproject.org:6969/ to see whichh ones are downloading the most of 15:14:19 Fedora-13-x86_64-DVD looks pretty popular. 15:14:22 by an order of magnitude 15:14:36 Fedora-13-i386-DVD same with that one 15:14:49 Can I get two +1s for this: http://dpaste.com/199000/ 15:15:00 +1 15:15:02 umm.. so even i'm confused on where to get the torrents 15:15:22 rather, i don't see the links anywhere on get-fedora 15:15:36 ricky: ^^ 15:16:34 Yeah, they aren't linked on fp.o at all, it was hard to find space with all the things we had to include :-( 15:16:50 gah 15:17:24 * ricky wishes we could just list 4 ISOs, desktop/DVD i686 and x86_64. 15:18:22 daMaestro: maybe make a note on the websites list 15:18:29 probably worth another discussion 15:18:45 * ricky is discussing with sijis in #fedora-websites 15:19:01 ah 15:19:15 He's in California, so times are a bit weird for him :-) 15:27:35 Alright, seeding both arch's -live and -DVD 15:27:37 * CodeBlock goes to get lunch 15:27:55 CodeBlock: thanks 15:28:14 i'm currently uploading at 2500KB/s 15:28:19 stupid box won't go any faster 15:39:39 mmcgrath: no problem, let me know if you come up with anything for me to do :) 15:40:43 dgilmore, jds2001: ping - mailman puppet config 15:40:59 we have manifests/services/mailman.pp and modules/mailman/ 15:41:11 and it sure looks like modules/mailman is NOT being used but the other is 15:41:29 so I'm just trying to figure out which one is the 'right' one 15:44:18 skvidal: i dont think we got it finshed ported to a module 15:44:33 okay 15:44:38 so my changes from yesterday are moot 15:44:41 I'll go fix them up 15:45:07 stickster, do you have more redirects you want me to do? 15:46:28 wow, best release day download experience ever 15:47:07 smooge: Still investigating 15:47:19 smooge: Weirdly, our docs-web repo appears to show that we *have* those CSS files 15:48:36 smooge: mmcgrath: http://git.fedorahosted.org/git/?p=docs/web.git;a=tree;f=public_html/es-ES/Common_Content/css;h=f2396150050d46d5e5167f3c9893ade3d99fe450;hb=master 15:50:57 * stickster wishes nb were around to ask 15:54:42 stickster: can you build your copy of that and send me the output of a "find" so I can see where all the files are? 15:55:46 mmcgrath: http://pfrields.fedorapeople.org/temp/css-list.txt 15:56:39 Hi 15:57:04 has someone ever looked at http://www.mirrorbrain.org/ 15:57:22 I think mdomsch has 15:57:23 gladiac, yes 15:57:31 I've spoken with peter poeml quite a bit 15:57:38 ah ok, fine 15:57:49 then I just hope Fedora will use it one day 15:57:52 MB is quite nice, MM accomplishes similar things with a different implementation 15:58:05 stickster: yeah, do a search on that page for "Release_Notes/Common_Content/" note that lang.css is gone 15:58:19 strange prediction :) 15:58:45 Oh bugger 15:58:54 mmcgrath: Good call. I'll put those in place and push them. 15:59:05 mmcgrath: A zero-byte file should be fine there, right? 15:59:33 stickster: it'd be even better if it didn't get requested at all but if that's a docbook convention that's hackish to fix a 0byte is fine too 15:59:46 unless that file should actually be there? 15:59:55 * mmcgrath isn't sure the mechanism that's actually creating the request and why 16:00:28 mmcgrath: It's to accommodate specific language CSS, i.e. RTL in some languages 16:01:06 When you build a document for publication in publican, it generates the default.css with those lines 16:01:35 16:01:36 It simply doesn't add a lang.css unless there's one it already knows it needs, like RTL for the he-IL (Hebrew) locale 16:01:43 so this could more be a publican bug? 16:01:48 mmcgrath: It definitely is 16:01:56 But we won't be able to get it fixed and pushed for at least most of a day 16:02:00 Since the maintainer is in BNE 16:02:21 So the fastest fix that the site's git pull won't overwrite (AFAIK) is putting in 0-byte lang.css for now 16:02:31 16:05:25 * stickster files Publican bug first to reference in the git changelog 16:07:14 mmcgrath: There's a good chance the Publican maintainer will say, "Fix your 404 redirects." 16:07:43 sites shouldn't have 404's anyway so even if we fix the redirects, we're still trying to serve content that doesn't exist. 16:09:04 * mmcgrath sees http codes 103 and wonders wtf thats about 16:09:40 103? 16:10:31 .bug 595799 16:10:32 stickster: Bug 595799 Missing lang.css file when publishing to web to avoid 404s - https://bugzilla.redhat.com/show_bug.cgi?id=595799 16:10:36 smooge: yeah 16:11:35 mmcgrath, you probably found this already http://jw.n--tree.net/blog/2009/9/23/http-status-code-103 16:11:39 mod cache 16:12:00 yeah it's cache related, I just don't get what is causing it and what the client might be seeing. 16:12:10 mmcgrath: :-( no access to push my change, apparently 16:12:38 ah hah, that's what's going on. 16:12:43 the wiki does something similar. 16:12:56 stickster: you can't push to the docs site? O_o 16:13:25 Make sure you get in docs-writers and docs-publishers 16:13:44 Looks like you're missing the second one 16:13:47 sparks: ^ 16:14:01 Yup, just saw that, fixing 16:22:32 If anybody's interested, here are spins stats for F12 (including both i686 and x86_64) 16:22:36 http://dpaste.com/199031/ 16:23:09 Do we have total download.fp.o stats for the DVDs and Desktop/KDE Live media? 16:24:16 /wiki/Statistics has combined stats: 1486587 unique IPs 16:25:07 those dammed LXDErs beating out Xfce! :) 16:27:55 Oxf13: By the way - we were talking about hardlinks and timestamp changes yesterday - if people sync with -H as suggested, then they shouldn't end up pulling more data 16:30:00 mmcgrath, which proxy is on ipv6? proxy4? 16:30:06 Yup 16:30:16 mdomsch: yeah, and shortly proxy2 16:30:28 Woah, PHX2 has ipv6? Cool 16:30:30 its response time is a lot slower than the others, at least for me 16:30:38 ricky: nope, it'll be in the UK 16:30:40 ricky, no 16:30:48 Ohh, that proxy2 :-) 16:31:04 $ time wget -6 -S -O - 'https://mirrors.fedoraproject.org/mirrorlist?repo=fedora-13&arch=x86_64&country=global' | wc -l 16:32:01 https://admin.fedoraproject.org/collectd/bin/index.cgi?hostname=proxy4&plugin=apache×pan=3600&action=show_selection&ok_button=OK 16:33:08 mdomsch: it's ipv6 specific, not proxy4 specific - 16:33:10 time wget -SO- --header="Host: mirrors.fedoraproject.org" 'http://152.46.7.222/mirrorlist?repo=fedora-13&arch=x86_64&country=global' | wc -l 16:33:19 ricky: the new proxy2 :) 16:33:27 mmcgrath, interesting... 16:33:50 dgilmore: ping 16:33:58 mdomsch: can you time this? 16:34:13 time wget -6 -S -O - 'http://fedoraproject.org/' | wc -l 16:35:22 mmcgrath: pong 16:35:25 mmcgrath, it's fine now - I restarted my aiccu tunnel 16:35:32 must have been a problem on my end 16:35:43 dgilmore: how's your ipv6 looking to Fedora services, particularly mirrors.fp.o 16:35:45 0.3 - 0.4sec now 16:35:48 mdomsch: interesting, lets get some verifications 16:35:55 please do 16:35:59 mdomsch: do you have header info from the -S portion of that request? 16:36:23 mmcgrath: ricky: OK, got my group access, now just need to wait for propagation 16:36:56 mmcgrath, http://fpaste.org/L8Q1/ 16:37:38 stickster: It's already synced to hosted 16:37:41 mmcgrath: ricky: OK, changes pushed, should be in next sync 16:38:13 mmcgrath: im getting 403's 16:38:30 mdomsch: do you have the headers from the slow request? 16:38:38 dgilmore: 403 to what address? 16:38:43 err URL 16:39:22 mmcgrath: i ran time wget -6 -S -O - 'https://mirrors.fedoraproject.org/mirrorlist?repo=fedora-13&arch=x86_64&country=global' | wc -l 16:40:07 Resolving mirrors.fedoraproject.org... 2610:28:200:1::fed0:2 16:40:07 Connecting to mirrors.fedoraproject.org|2610:28:200:1::fed0:2|:443... failed: Permission denied. 16:40:49 permission denied to a 443 port? 16:40:51 that's... weird 16:41:03 i can ping6 the host ok 16:41:21 dgilmore: wget -6 -SO/dev/null http://fedoraproject.org/ 16:41:22 try that 16:41:37 dgilmore: what headers did you get? 16:41:42 an interesting question here, but is it intended that no torrent options are available on get.fedoraproject.org? 16:41:49 permission denied 16:41:58 I can't help share if I can't find the links... 16:41:59 herlo: I'm not sure what the deal is with that, check in #fedora-websites I know they've been discussing it. 16:42:06 mmcgrath: cool, tx 16:42:13 http://torrents.fedoraproject.org/ ;) 16:42:28 (which is where we have been pointing folks in #fedora who can't find them) 16:42:37 It's intended 16:42:50 We discussed this with mizmo, and decided to leave it as is 16:42:53 so are we going to slowly phase out torrents? what's the plan there? 16:42:56 mmcgrath: i think its something with me 16:43:02 nirik: but that's not my point :) 16:43:06 dgilmore: selinux? :) 16:43:07 mmcgrath: i cant access ipv6.google.com either 16:43:14 * ricky personally wanted torrents to be linked somewhere, but ah well. 16:43:24 mmcgrath: selinux is in permissive 16:43:31 herlo: I'm just saying we are seeing that in #fedora a lot too... ;) 16:43:34 ip6tables? 16:43:34 ricky: I think I have an easy solution for the future and could hack it in quickly... 16:43:35 since my kvm setup doesnt work in enforcing mode 16:48:29 dgilmore, I had to restart my sixxs tunnel 16:49:04 mdomsch: i did that 16:49:24 mdomsch: its something on my external router box 16:49:36 i run aiccu on my colo'd box 16:49:45 and have a tunnel from home to it 16:52:29 i think its actually something outside of my control 16:53:43 ssh over ipv6 is working 16:53:48 but http and https is not 16:54:03 dgilmore: not ipv6tables? 16:54:30 mmcgrath: no my router box has no rules blocking forwarding packets 17:07:12 One question I just migrated all my mailing list subscription to gmail and do I need to ping each list admin to accept my new address be able to reply and post to the lists ? 17:07:28 ricky: it'll still cause some shuffle and disk I/O on the side of the mirrors, removing existing files and replacing them with hardlinks. 17:07:36 warthog9, how's your traffic look? 17:09:07 Viking-Ice, most fedora lists don't require admin approval to join 17:09:38 stickster: just curious - what's in 'oldguides' under docs.fp.o/ 17:10:39 also - is there a reason why our docs.fp.o subdir on the proxy servers comes with a .git subdir that's 2.9GB? 17:10:41 mdomsch: After the change I'm unable to post to the list ( receiving is fine ) mailing address being blocked for some reason ? 17:12:12 nb: ping 17:12:42 nb: so the docs.fp.o subdir that's on all the proxy servers - it's 6.6gb and it has 2 troubling items 17:13:01 nb: 1. it has a .git dir that's not being excluded so 2.9GB of files no one will ever see are in the dir 17:13:19 2. oldguides appears to have all the former docs? for releases going back to f8? 17:13:23 which seems..... odd 17:13:51 skvidal: git clone http://docs.fedoraproject.org/ ? :-) 17:14:04 mdomsch: it's replicated to all the proxy servers 17:14:18 so each one of them is carting around 2.9GB they don't need 17:14:20 skvidal: I'm probably not the best person to ask about how that repo works -- nb set it up for Docs 17:14:20 right? 17:14:22 skvidal, agreed 17:15:04 As far as oldguides goes, I *think* rudi is working on making them available through the site menu -- we previously had them on the old docs.fp.o site as well 17:15:56 okay 17:16:12 they're 1.4GB and maybe it wouldn't be rude to nuke the f10 and below versions? 17:16:24 skvidal: I think f10 is on the menu already 17:16:27 http://docs.fedoraproject.org 17:16:56 why? 17:17:20 mdomsch: ignore-it post present in list archive client problem.. 17:17:38 skvidal: ----> docs@lists.fp.o 17:17:46 righty-o 17:17:47 Oxf13: Sure. I guess you could ask around on mirror-list-d and see what they think 17:18:01 yeah, I'm sure they'll be for it, once we give them a heads up 17:18:03 just not during release time 17:18:12 Oh yeah 17:18:19 skvidal: I truly don't know, I asked about carting around old docs a year or so ago, and the team said they wanted to retain old docs 17:18:32 skvidal: However, the new site breaks the old links anyway, so the value's somewhat lessened of late 17:20:05 mirror.its.uidaho.edu is getting about 100Mb of traffic, I can handle at least 200Mb commercial plus up to a Gb of I2/NLR. Feel free to send me twice as much traffic, if needed, and if possible. 17:23:25 * stickster is wondering where the Moblin spin went 17:24:01 davelien, I've bumped your bandwidth value in MM. watch it after the top of the next hour 17:24:59 mdomsch: can mirror admins control that value? 17:25:09 Jeff_S, yes 17:25:23 it is nominally the bandwidth in Mbits/sec you can serve 17:25:37 is that the same as the number shown on the publiclist page? 17:25:40 but it simply feeds the weighted random draw 17:25:41 yes 17:25:50 oh, ours is quite low there, I should change it 17:26:43 If I remember correct the only choices on the admin page are 100Mb or 1Gb, has this changed? 17:26:53 davelien, it's just a number now 17:27:10 davelien: also, thanks for your mirror, I use it occasionally for stuff we don't have ;) 17:29:15 Ah, right you are Jeff_S, just checked to page. I'm set to 600Mb so I guess I'll just let the weighted random draw algorithm do its thing. 17:29:27 Jeff_S, you are welcome! 17:29:32 davelien: I just do what mdomsch tells me ;) 17:29:56 I just bumped up our number to 500 to see what happens 17:30:53 davelien, I just bumped you from 300 to 600 17:32:47 Oxf13: We appear to be missing a spin -- Moblin's gone MIA 17:34:12 stickster: not listed at https://fedoraproject.org/wiki/Releases/13/Spins 17:34:58 Oxf13: https://fedoraproject.org/wiki/Spins_Process says Spins SIG is supposed to notify people 17:35:53 stickster: yes, the spin sig doesn't really work 17:35:59 * nirik is sorry if something broke down. I thought moblin was approved, but then in limbo, then ok. 17:35:59 stickster: I'm surprised you haven't noticed. 17:36:23 The last list I saw from the Spins SIG was the notification here: http://lists.fedoraproject.org/pipermail/rel-eng/2010-February/009387.html 17:37:04 i thought moblin was on hold due to upstream going a little odd in the meego merge/rebranding 17:37:30 I wish it did... I'm not sure I have time or energy to make it work, but I guess if noone else appears I should make time to do so next cycle. 17:37:31 notting: It's not on hold, but I know Peter is looking to move to MeeGo after they have a genuine release 17:38:14 Board approved Moblin in January: http://lists.fedoraproject.org/pipermail/advisory-board/2010-January/007869.html 17:38:16 stickster: but i thought what was in rawhide was a pre-release of a moblin-that-will-never-be-released-as-moblin software set. or something. 17:38:30 i could be misremembering 17:38:33 notting: yes, but then they did finally release 2.1 17:38:37 (I think) 17:38:50 They did 17:39:43 nobody updated the wiki page, which is what I was told to use as the definitive list of what to compose 17:39:54 and nobody mentioned anything missing when I posted to the spins list asking maintainers to review 17:40:09 so no moblin for F13 17:40:11 process fail. 17:40:37 jwb, Oxf13: missing signatures from the latest push? 17:40:47 How can we get the Moblin spin out there, since we've told people in many places it's available? 17:40:52 lmacken_, no? 17:41:06 jwb: ok... well, something fucked up :) 17:41:10 stickster: do you really want to push a completely untested iso out there? 17:41:25 lmacken_, oh that. koji just pruned signed copies. it happens a lot 17:41:27 i'll fix it 17:41:40 jwb: thanks 17:41:58 stickster: were you/people not working off of https://fedoraproject.org/wiki/Releases/13/Spins ? 17:42:31 notting: Apparently not everyone was -- I got an email from pbrobinson asking where the Moblin spin was 17:43:07 I just read the Spins process page and it says that wiki page "should be updated" -- note the passive voice unfortunately doesn't assign any responsibility for doing it. 17:44:15 stickster: spins wrangler ? 17:44:20 given that probinson didn't say anything when I asked the spin sig members to test their spins two weeks ago, I'm less than eager to scramble to rush something out. 17:44:45 the person who was managing spins seems to have walked away from it, without a lot of notice (or without me noticing it much other than the lack of work) 17:45:11 spins: not one of our better examples of "value" 17:45:32 spins: we want it all, but we don't want to do any work for it 17:45:58 bruno and I are somewhat around for spins, but we are both busy with other stuff. Most of the other spins maintainers don't provide any input at alal. 17:46:01 jwb: it's the 5th f 17:46:12 ? 17:46:18 freedom 17:46:18 ail 17:46:20 friends 17:46:22 first 17:46:24 fucked 17:46:24 nirik: yeah, I think you've gone above and beyond 17:46:26 oh hmm 17:46:28 fail 17:46:28 spins weren't your bag to carry 17:46:30 yah that works 17:46:45 Oxf13: I made a choice to try and help it out, but not run it... just ENOTIME 17:46:48 * skvidal is kidding 17:47:07 if the biggest mess up is miscommunication on one of the spins then that's looking awfully good :) 17:47:15 nirik: I made a choice specifically not not jump in and mop it up. I didn't want to be left holding the bag. 17:47:43 skvidal: it's really just the biggest example of the process breakdown, or the most visible. 17:47:46 frustration, fucked, fail, finicky, flagellation, fap, fubar 17:47:49 skvidal: Oxf13: That's fair -- Spin owners have to understand the process. I don't think it changed between F11->F13,did it? 17:47:54 all good choices 17:48:03 stickster: nope... should be all the same from what I recall. 17:48:15 stickster: not really, not that there has ever been a workable process 17:48:30 there really hasn't 17:48:37 * stickster feels bad that pbrobinson is left in the lurch again. This happened with Moblin before. 17:48:39 * Oxf13 notes that we list Haskell on the wiki page, but there wasn't a KS for it to make 17:48:51 I think it could be better, but there is a process. 17:48:58 stickster: then he should have been motivated to make sure everything was going according to plan 17:49:03 weeks ago, when I asked for people to. 17:49:08 Haskell is a ks only spin... shouldn't be produced, just added to spins-kickstarts. 17:49:23 there IS a spins process, but not always full attention to detail in following it or managing it 17:49:24 nirik: yeah, but then why i sit listed at https://fedoraproject.org/wiki/Releases/13/Spins ? 17:49:45 Oxf13: to advertise it's existance in f13? 17:50:10 Here is this thing which we have nothing to give you... 17:50:22 a ks file? 17:50:30 (sorry, just being snarky here) 17:50:33 it's our new plan to attract contributors 17:50:47 WANT THIS COOL THING? GOOD! MAKE IT YOURSELF! 17:52:42 mmcgrath, ping 17:52:48 The process needs revisiting, and a leader/etc needs to exist. 17:55:05 * nirik moves on from this relaxing and pleasent conversation to prep for the fesco meeting today. More fun. ;) 17:55:09 Heh 17:56:10 nirik: I could probably arrange to make fesco equally as relaxing and pleasant, if you'd like 17:56:18 nirik: I'm sure I could rile kofler up 17:56:38 the challenge is to not do so. ;) 17:58:11 nirik: nah 17:58:14 skvidal, too late 17:58:18 * ricky goes off for a while, see you later 17:58:21 mizmo: :) 17:59:40 notting: Aha. http://lists.fedoraproject.org/pipermail/spins/2010-April/001081.html 17:59:56 I was wondering about where you thought you had seen an on-hold status. 18:00:38 spoleeba: ping 18:00:40 spoleeba: err pong 18:00:54 spins are bad for my blood pressure 18:01:18 mmcgrath, got time for a PM 18:01:25 of course, hit me! 18:01:33 * stickster sees that Peter did respond to testing call here: http://lists.fedoraproject.org/pipermail/spins/2010-May/001089.html 18:09:34 stickster: that was the may 5th call, I made another on May 11th with actual produced isos to test 18:22:48 Oxf13: ping, can you check if that crazy .nfs file is still there? 18:23:00 * mmcgrath already forgot the path 18:23:08 oh I guess it's in the ticket I created though 18:23:09 * mmcgrath checks 18:24:24 Oxf13: yeah it's gone now. you can try to remove whatever again. 18:25:39 Oxf13: if you have a moment, can you try to recreate that permission denied by group error? 18:26:58 nb: ping? 18:32:19 mdomsch: looking 18:32:46 mdomsch: broken ;-) 18:32:50 (looking at it now 18:39:28 cache is taking up more room on proxy06. I would like to delete the .git in /srv/web/docs.fedoraproject.org before we run out 18:39:57 smooge: Id' +1 that 18:40:14 can I get another? 18:40:31 mmcgrath, ricky ? 18:40:37 smooge: it'd get put back in place on the next rsync run 18:40:43 and eat up bandwidth doing so :) 18:40:53 where does it get rsync'd from 18:42:11 smooge: all you really need to do is add an exclude to the sync 18:43:04 smooge: bapp1 18:43:17 this is likely just a simple fix in the build script (which happens on bapp1) 18:43:23 nb would know for sure but I think he's been busy today. 18:43:38 is trac the right place to report an issue with hardware detection tool in BFO? 18:43:40 ok I will start digging 18:43:58 mdomsch: of course 2 weeks ago cacti broke and I didn't notice 18:44:08 gui1ty: no the HDT should be reported to syslinux 18:44:19 I think we have another 2 hours before we run out on proxy6 18:44:26 gui1ty: I guess it depends on the issue, if it's HDT itself probably upstream to syslinux. If it's a problem loading it or something that might come to us 18:44:27 gui1ty: or to BKO 18:44:47 smooge: huh? how's that? 18:44:53 is the docs site growing over time? 18:44:59 mmcgrath, no cache is 18:45:16 lets watch it during that time, cache should even out and get pruned. 18:45:19 warthog9: directly or in bz (for syslinux)? 18:45:35 gui1ty: what is the problem btw? :) 18:45:52 the pci.ids file is missing or not in the right place 18:46:24 gui1ty: oh, that could very well be me, give me a bit to take a look. 18:46:29 gui1ty: ohhh if it's an issue of the pci.ids file not in the right place that's mmcgrath's problem ;-) 18:46:40 warthog9: thanx ;-) 18:46:48 * warthog9 notes that mmcgrath clearly didn't copy that bit out of BKO 18:47:31 gui1ty: your IP start with 80.101 ? 18:47:57 warthog9: heh, our HDT is both useless and funny at the moment :) 18:48:48 mmcgrath: how is it funny? 18:49:22 gui1ty: it's just all unknown. I should have that fixed in about 2 minutes. 18:49:52 * gui1ty is realizing that he downloaded the BFO images a while ago 18:50:04 mmcgrath: in what way? 18:50:24 warthog9: it's just that since the pci.ids file is missing HDT can't actually identify anything and lists them all as unknown:unknown 18:50:32 mmcgrath: since it's still there I'll wait for the fix 18:51:08 mmcgrath: the numeric ids are shown on the bottom line, so it's not all useless ;-) 18:51:58 gui1ty: ok, try it now, you'll likely have to re-load the HDT 18:54:28 mmcgrath: now it takes some more time to load and all the device descriptions are there 18:54:32 mmcgrath: thanks 18:57:13 mdomsch: apparently *mysql* broke a table that cacti needed 18:57:21 warthog9, whoops 18:57:27 mdomsch: yeah 18:59:55 warthog9: All part of the sekrit oracle agenda ;-) 19:00:41 abadger1999: figures 19:00:53 abadger1999: I flip everything over to mysql and *THEN* things start breaking 19:01:30 hehe 19:03:01 figures 19:04:28 looks like both my cacti instances had the same mysql problem, that's a little concerning 19:05:11 warthog9: FYI... there's a recent cacti vulnerability out. The fix has been pushed, it'd be 0.8.7f 19:05:31 mmcgrath: joy 19:05:42 mmcgrath: have to see what it is / how bad 19:06:29 warthog9: https://bugzilla.redhat.com/595289 19:06:58 mmcgrath: ok that pushed out today 19:08:02 mdomsch: there we go new data starting to flow in 19:08:24 warthog9, url? 19:08:33 http://cacti.kernel.org/graph_view.php?action=tree&tree_id=1 19:08:45 'default tree' on the left it it doesn't take you right there 19:09:03 might take it another 15 minutes to have something actually interesting though 19:09:22 running iptraf on one of the boxes now 19:09:27 does fedora run cacti? 19:09:33 geppetto: After reading some of your yum commits I think I've got your hashing gotcha documented: https://fedorahosted.org/releases/k/i/kitchen/docs/api-text.html#unicode-str-and-dict-keys 19:09:39 jokajak: we used to, not anymore though 19:09:51 running something else now? 19:10:00 jokajak: collectd - http://admin.fedoraproject.org/collectd/ 19:10:04 mdomsch: if I'm reading this right I'm not moving a lot of US traffic (though getting hit relatively hard on the load) 19:11:17 mdomsch: I'm only moving ~300mbps on zeus1 19:12:01 someone: is there a procedure (can't find one) in case the reporter of a review request hasn't responded to his review request for a long period of time? 19:12:09 it's so ugly :( 19:12:38 mdomsch: and if the aggregate bandwidth graph is right (looks like it) I'm moving less then 2gbps worldwide, which is barely above normal background 19:12:41 wzzrd: I think so, check in #fedora-devel, they'd know. 19:12:47 warthog9: :( 19:12:59 mmcgrath: they just pointed me here :D 19:13:39 why did they do that? 19:13:47 wzzrd: It's a #fedora-devel question but the answer is that you can ping and say you'll close it in X days (or take over the submission of the package in X days if you want it in the distro). 19:13:59 mdomsch: I suppose the question is are you actually redirecting a proportionate amount of traffic to me, or is this release a dud? 19:14:22 should still be proportionate 19:14:25 mmcgrath: abadger1999: ah! just got an answer in #fd; but thanks anyway 19:15:55 warthog9, the other 3 are north of 500mbps 19:16:37 mdomsch: yeah, but that's a far cry from the 3.7gbps peak I've seen from a Fedora release 19:20:01 F13 is bad luck, people are waiting for F14 :) 19:20:35 cemc: maybe ;-) 19:21:16 cemc: though reading some quick upgrade notes for F13, things like no longer being able to disable KMS for intel might be causing some pause 19:21:46 abadger1999: That's mostly it ... but the biggest problem we had was that ("a", "b") == (u"a", u"b"), with hashes too 19:22:08 geppetto: Okay -- so you'd like a note of this applying to tuples as well? 19:22:50 abadger1999: Yeh, might as well put a small bit in there ... saying "note hashes of tuples are hashes of their members" 19:22:54 abadger1999: or something 19:23:57 abadger1999: Or maybe something like: "all dict keys should be the same type, is a good general rule but note that breaks down when the type is tuple" 19:30:36 skvidal, /usr/bin/rsync -aSHP --exclude=.git bapp01::docs/ /srv/web/docs.fedoraproject.org/ should clean things up nicely 19:30:49 well 19:30:54 except it doesn't delete, does it? 19:31:06 skvidal: we can do that manually. 19:31:10 okie doke 19:31:11 smooge: you going to put that change in? +1 from me 19:31:18 * skvidal is also +1 19:31:25 mmcgrath, I am needing to check with something 19:32:00 the tree public_html/en-US/Fedora_Contributor_Documentation/1/html/Publican_1.6_Users_Guide is on the proxy servers but not on bapp1 anymore 19:32:23 I am trying to find out why 19:33:02 smooge: looks like the rsync call being used doesn't have a --delete ? 19:33:22 no it doesn't 19:33:37 so anything over the last X time will have gotten on there. 19:33:39 * warthog9 debates how much he's going to hate f13 by upgrading his primary home fileserver / server to f13 ;-) 19:33:44 Doing the --delete cleans up soooo much 19:34:07 * jokajak votes rhel6 19:34:23 I am going to commit the change in a short while. 19:34:33 f13/rhel6 - potayto/potahto :) 19:34:58 jokajak: rhel's fine until your 3 years down the road and you suddenly start realizing how old it is in comparison to things like Fedora 19:35:38 psh, what more could you need from F12 than what's there? ;-) 19:36:56 warthog9, yeah but 3 years down the road you are supposed to throw away that computer and get something new 19:37:21 smooge: you clearly have more budget then I do for fancy new hardware ;-) 19:37:24 smooge: hell, 3yrs down the road you HAVE to throw it away 19:37:34 smooge: I've not had a laptop that actually survives 3yrs of use 19:37:37 smooge: just think I still have CRTs :-P 19:37:59 the only lcd's I have are on laptops or my TV 19:38:09 my home router is a laptop that is 4-5 years old 19:38:25 * abadger1999 even has a crt hooked up to a laptop... 19:38:44 * notting has a CRT monitor sitting on the floor he hasn't disposed of yet (and two CRT TVs) 19:38:49 warthog9, well every 3 years I go to the used computer store to buy something newer. I am just about ready to get rid of my Dell Optiplex from 2001 19:39:01 smooge: lol 19:39:10 smooge: yeah my mythtv setup runs Fedora 7 19:39:17 built maybe a year before that came out 19:40:26 I am hoping to ressurect my wife's Apple Powermac from 2002 to have something a bit faster. 19:40:56 The kid has faster newer computers than me... his XO is probably faster than the Optiplex 19:43:17 smooge: hehe my phone now ranks in the faster machines in my apartment 19:49:13 Ticket notification - infrastructure: Ticket #2187 (enhancement closed): Stopgap fixes for Docs site on release day 19:54:16 mmcgrath: you around? 19:59:11 lmacken: Do you have a group for the critpath -setting script to belong to? 19:59:36 abadger1999: we currently don't have a user or group for setting that 19:59:52 nirik: yup, whats up? 19:59:59 lmacken: Think of a couple names :-) 20:00:11 mmcgrath: FES ticket roundup in the fesco meeting right now. ;) 20:00:56 abadger1999: masher? critpathbot? 20:01:05 nirik: ah 20:01:44 lmacken: k I'm adding user: critpathbot group: critpathadmin to fas now. 20:02:06 abadger1999: sounds good to me, thanks. 20:02:14 mmcgrath: wiped out 13.RC3, thanks. 20:02:21 mmcgrath: ok, check out spins/releng/test/ 20:02:53 $ touch spins/releng/test/foo 20:02:54 touch: cannot touch `spins/releng/test/foo': Permission denied 20:06:40 Oxf13: one sec, poking 20:08:19 Oxf13: I'm going to become you and run a couple of commands for this ticket. just a groups and a touch, cool? 20:12:19 sure 20:14:43 rut-roh 20:14:44 abadger1999: ping 20:14:52 mmcgrath: pong 20:15:02 abadger1999: hey, got a problem on cvs1 - 20:15:14 ps auxwww | grep -c toshio 20:15:14 mmcgrath: what's up? 20:15:30 looks like genacls is hanging 20:15:35 some of these are from yesterday 20:15:37 holey mother of god. 20:16:12 nice 20:16:18 another need for a wrapper! 20:16:24 I'm going to start by killing all of those and then we'll see if a new run works. 20:20:54 mmcgrath: Okay, seems to be cvs that's hanging on commit. 20:21:15 ah, probably a lock file? 20:23:39 mmcgrath: Yep, lock file -- all set now. 20:24:06 skvidal: I removed a stale lock file for your user on cvs01 on CVSROOT/admin 20:24:12 skvidal: Was created yesterday. 20:24:23 abadger1999: yesterday. 20:24:25 abadger1999: oh 20:24:29 I ctrl-c'd something 20:24:31 abadger1999: thanks 20:24:36 k. No problem. 20:36:44 .fasinfo critpathbot 20:36:45 mmcgrath: User: critpathbot, Name: Critpath setter in Pkgdb, email: toshio+critpath@fedoraproject.org, Creation: 2010-05-25, IRC Nick: None, Timezone: UTC, Locale: C, Extension: 59923, GPG key ID: None, Status: bot 20:36:49 abadger1999: ping 20:36:49 mmcgrath: Approved Groups: critpathadmin 20:36:55 mmcgrath: pong 20:37:04 mmcgrath: I created that jsut now... what's up? 20:37:23 oh, nevermind. so you did. I just got an alarming email about it but now am less alarmed :) 20:37:30 Warning: Someone attempted to reset the password for system account 20:37:49 Ah -- right. I forgot that we disabled changing that via the web. 20:39:18 mmcgrath: i just saw that 21:03:17 K, looks like things are moving back down to normal - http://mmcgrath.fedorapeople.org/monitoring/proxy-requests.html 21:03:49 mmcgrath: Did the 0-byte link.css entries help relieve some of the problems? 21:04:40 stickster: it did, the next biggest offender is this one - 21:04:44 skvidal: Question -- is it possible to retain .git/refs/heads/* in the proxy copies of docs.fp.o, so we can identify where the current copy's drawn from? 21:04:49 File does not exist: /srv/web/docs.fedoraproject.org/public_html/images/title.png, referer: http://docs.fedoraproject.org/en-US/toc.html 21:05:22 stickster: I guess- you'd need a kinda zany include/exclude to rsync but do-able 21:05:32 stickster: might be easier to put some version-file in the dir itself 21:05:35 outside of git 21:06:02 skvidal: Ah, it's using rsync 21:06:15 yes 21:07:09 so: --include '.git/refs/heads/**' should be OK 21:07:21 or something like that, I haven't seen the rsync command in question 21:07:24 nod 21:07:30 * skvidal points at smooge 21:07:37 * stickster tips hat :-) 21:08:51 what huh 21:09:02 did I do something 21:10:00 stickster, the rsync command is currently: /usr/bin/rsync -aSHP --exclude=.git bapp01::docs/ /srv/web/docs.fedoraproject.org/ 21:14:53 stickster, also we are still seeing css problems 21:18:06 smooge, css issues? 21:18:11 smooge: So the above should work: --include .git/refs/heads/** 21:18:19 (after the --exclude I imagine) 21:18:30 smooge: As for CSS -- ugh, what's going on now? 21:18:31 * mizmo offers css assistance if its helpful 21:18:33 smooge: did you notice your rsync commands produce output? see admin@fp.o 21:18:53 I think you need a -q 21:19:06 mizmo: I think this issue -- if it's related to an earlier one -- isn't design, it's more missing files on the docs.fp.o server that are causing 404's and load 21:19:28 smooge: Come by #fedora-docs and we can explore. Rudi is up now and can help while I'm away 21:19:37 stickster, mmcgrath http://fpaste.org/aqhH/ First column is the missing .css and the second is the file that is asking for it 21:19:47 smooge: thanks 21:20:49 ah okay 21:20:51 smooge: Is it the other way around? 21:21:05 * stickster sees some have a dir first, a CSS file second 21:21:22 geppetto: http://fpaste.org/XVxn/ 21:22:38 geppetto: Does that work? If so I'll include it in lieu of my other example 21:22:59 stickster, it is an awk from the line say "File not found: /srv/web/docs/spew/, referrer http://spew/ 21:23:11 abadger1999: wow 21:23:18 smooge: Ah, so it could be a broken link in old docs 21:23:37 abadger1999: that might actually work, and solve our intern strings problem in yum :) 21:24:02 geppetto: hehe if so, I'm glad to have helped :-) 21:25:47 if I turn that into a dict replacement class, it could go into kitchen directly.... /me works on that. 21:38:38 abadger1999: It seems to work here, and saves about 2% over just ignoring tuples ... alas. the CPU time goes up by 50% 21:38:59 geppetto: :-( Yeah, iteration sucks the life out of things. 21:39:31 * geppetto nods 21:40:31 geppetto: If you have enough data to benchmark.... something to try tweaking would be with and without the hash() check. 21:41:04 Since you know you're only passing hashable values in the first place, that might be a win. 21:41:11 abadger1999: I dropped the hash check :) 21:41:15 Okay :-) 21:43:02 abadger1999: I think one problem is that for a smallish "yum update" we are calling type_code 115,275 times :) 21:43:24 That's a bottleneck... function calls are expensive. 21:45:02 * abadger1999 can't shake the feeling if he were just a little smarter he could do this without iteration. 21:47:28 Oxf13: can you approve the email i just sent to devel-announce 21:48:41 dgilmore: waiting for it to show up 21:49:06 abadger1999: I've tried a couple of hacks, I can get it down to 30% CPU for 3.5% RAM ... which is still terrible 21:49:30 geppetto: Were your hacks replacing recursion with iteration? 21:49:33 One of these days, probably soon, we'll probably have to make sure all pkgtups are in bytes anyway 21:49:42 21:49:45 Oxf13: hrrm i got a message saying "Posts are allowed only from subscribed members." 21:49:54 Oxf13: i wonder if it just got dropped 21:50:01 must have. 21:50:02 geppetto: yeah, for yum and rpm that seems like a necessity. 21:50:27 * geppetto nods ... soooo much pain though :) 21:51:49 * geppetto goes to find some food 21:53:12 geppetto: Ohh... What about using repr? 21:53:23 as in store[repr(key)] = value 22:01:45 abadger1999, ricky mmcgrath skvidal I need an approval to push a redirect update 22:07:08 smooge: So does moving the accessibility guide and deployment guide do anything? 22:07:41 smooge: ie: the only change that changes what users see is the release-notes part? 22:07:50 no it was just putting them in alphabetical order so I could keep track of them 22:07:58 Gotcha. +1 22:07:59 I am expecting more 22:08:27 I'll put that in email for the record keepers. 22:37:06 Ticket notification - infrastructure: Ticket #2189 (enhancement created): FAS plugin for zikula should prefill real name for users  || Ticket #2188 (enhancement created): Fedora Events "space" in Fedora Insight 22:38:38 hello, we are a mirror that just discovered we are pointed to the wrong rsync host. what is the proper rsync host? 22:41:23 pashdown: http://fedoraproject.org/wiki/Infrastructure/Mirroring/Tiering 22:56:19 hello, someone in #fedora referred me here due to my question about the new boot.fedoraproject.org install method, it appears the conf file hasn't been updated with the 13 release info: http://alt.fedoraproject.org/pub/alt/bfo/pxelinux.cfg/fedora_install.conf 22:59:57 droidix: how'd you get directed there? that's not the official bfo location - http://boot.fedoraproject.org/ 23:00:29 * mmcgrath goes to update that site in case anyone else finds it there 23:00:50 that was the file the installer downloaded when booting 23:01:38 droidix: when did you download it? 23:02:09 I downloaded the USB image at 10:51 CDT this morning 23:02:19 interesting, looks like I've got some updating to do. 23:02:24 droidix: sorry for the trouble, it should be fine now. 23:02:37 droidix: thanks for the heds up 23:02:40 err heads even 23:02:40 ok, no problem, thanks! I think this is a really sweet install method 23:02:44 * mmcgrath bbiab 23:06:47 mmcgrath, hmm how does anyone know about bfo? Its still sort of tucked away right now isnt it? Or did it make the release notes? 23:07:09 yes, it was on page 2 of the release notes :) 23:07:20 that's how I saw it, pretty prominent position 23:07:32 http://docs.fedoraproject.org/en-US/Fedora/13/html/Release_Notes/sect-Release_Notes-Installation_Notes.html 23:07:36 droidix, not exposed on getfedora however 23:07:49 Two clicks off the fedora home page 23:07:58 droidix, im more impressed that you read the release notes 23:08:33 spoleeba: it was in the announce email also 23:09:15 lol, thanks, I really do think this method could revolutionize installs 23:35:42 mmcgrath: what's up 23:37:46 CodeBlock: hey, not much here. building night stands. 23:38:12 CodeBlock: how's your pt server working out? 23:39:09 mmcgrath: dgilmore gave me access to pt3 so I was able to test groups in the script, seems to work fine 23:40:01 good 23:40:09 mmcgrath, smooge skvidal did you get .git excluded in the rsync cron? if not i can do it 23:40:12 #info so far so good. I'm going to call this a successful release. 23:40:15 #endmeeting