21:00:02 #startmeeting EPEL SIG 21:00:02 Meeting started Fri Sep 11 21:00:02 2009 UTC. The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:02 Useful Commands: #action #agreed #halp #info #idea #link #topic. 21:00:09 #topic roll call 21:00:55 here 21:01:00 for epel meeting 21:01:04 welcome. :) 21:01:13 * stahnma is in 21:01:41 * mmcgrath is here 21:01:53 smooge / jds2001 / rayvd ? you guys around? 21:03:14 ok, lets go ahead and get started I guess... . 21:03:20 #topic Status update on action items 21:03:36 stahnma: anything to report from the summit? and/or inode0? 21:04:05 * warren watching 21:04:40 not a whole lot 21:05:00 from the few I spoke to about EPEL, most don't use it much 21:05:05 they grab RPMs from wherever 21:05:21 and mostly deal with things like oracle, db2, weblogic and crap like that 21:05:24 * mmcgrath thinks we need marketing :) 21:05:24 :) 21:05:41 * stahnma thinks we need solid practices, oh and the packages people need 21:05:43 yeah, marketing would be nice. 21:05:56 that too. 21:05:58 stahnma: ubuntu has proven we don't need that to be successful :-P 21:06:20 sorry I'm late 21:06:27 no worries... welcome maxamillion 21:07:00 should we look at doing talks/something at summits? 21:07:22 nirik: talking about FUDCon or $other? 21:07:26 nirik: maybe, but so few people actually end up going to the summit 21:07:33 i am here. 21:07:35 and the FUDCon people know about it 21:07:37 sorta. while at work. :) 21:07:38 maxamillion: Red Hat summit 21:07:43 no need to advertise to ourselves :) 21:07:47 good point :) 21:07:57 nirik: ah ... yeah, wanna go to the next one ... I'd love to give a talk if they'd let us 21:08:34 yeah, I don't know what kind of crowd is at the summits... the only one I went to was the one with a fudcon attached. 21:09:07 I would have liked fudcon to be attached 21:09:11 any idea how many epel users there are out there? 21:09:23 preaching community without Fedora there kind of defeats itself 21:09:30 derks: no idea. ;) we could ask mmcgrath to look at some mirror stats for us or something... 21:09:32 'users' is a tough thing to track if it means the number of humans that use it 21:09:54 i guess i mean boxes subscribed... which is hard to pull from yum 21:10:00 and a hundred mirrors 21:10:27 maybe if you look at the stats for mirrormanager | grep epel 21:10:52 err mirrorlist 21:11:15 there are not nearly as many epel mirrors as fedora mirrors... 21:11:32 but it's still really hard to say how many end machines/users there are. 21:11:50 anything more on the summit? or shall we move along? 21:11:57 looking at just yesterday we had 49,184 unique IP's and 163,220 individual hits. 21:12:04 which to me implies lots of nat'ing 21:12:30 I think that's a decent number of users 21:12:36 and there are several internal mirror of epel 21:12:37 nirik sorry was in another meeting 21:12:39 yeah, I would expect more in the epel world of that than in the fedora world. 21:13:16 no doubt, I'd say we're healthy if nothing else. 21:14:03 so I guess the question is, would the Summit be accepting of EPEL talks? 21:14:22 its definitely growing via our customer base, but there is still a stigma around using non rhel repos 21:14:28 maxamillion, thats up to people outside of our control :) 21:14:41 nirik: it was roughly 15% of the total unique IP's that hit our mirrorlist yesterday 21:14:54 cool. 21:14:54 wow, that's pretty good 21:15:01 smooge: right, which is kinda of my point .... do we want to discuss topics and points of interest before we have a "yes" or "no"? 21:15:13 yes 21:15:32 closer to 17.5% of the total hits to the site. Further evidence of more nat in EPEL then Fedora :) 21:15:51 because you have to have topics of interest before you can find out what will be accepted 21:16:01 smooge: fair enough 21:16:55 mmcgrath: i expect that epel mirror stats are heavily underestimating useage 21:17:17 mmcgrath: since most people using rhel/CentOs are like to have multiple machines behind nat 21:17:23 dgilmore: me too, I bet there's lots of people using EPEL that have resources (and reason) to run their own mirrors as well. 21:17:32 mmcgrath: right 21:18:37 good topic for a potential talk would be telling people how to pull epel in as a custom channel to their satellite server 21:18:54 I think the spacewalk list has a script for that 21:19:05 rayvd++ 21:19:05 if not, they have something very similar 21:19:16 perfect. 21:19:24 that would be great (rh satellite admin here) 21:19:35 or convince RH to make it available in RHN :P 21:19:57 nirik: im semihere 21:19:58 rayvd: that would be amazing ... though I imagine highly unlikely 21:20:14 yeah, probably. :) 21:20:23 in spacewalk (currently) you can import a yum repo 21:20:26 and it reposyncs it 21:20:35 satellite 5.3 might have that (not sure) 21:20:36 stahnma: i was just gonna say that :) 21:20:43 if not, 5.4 maybe :) 21:20:48 nice. i know our version of sat doesn't do that, but you can do other things to pull the rpm's in :) 21:21:23 stahnma: i dont think its in satellite 5.3 it was new in spacewalk 0.6 21:21:36 stahnma: and satellite 5.3 is based on spacewalk 0.5 21:21:52 ah, ok 21:22:28 something explaining how to do that on the wiki would be cool. 21:23:26 anything more on the summit? or shall we move on to other action items? 21:23:34 nothing really more from me 21:24:14 Next item: jds2001: Incompatible version upgrades process/guidelines 21:24:18 any news on that? 21:24:52 * jds2001 sucks. 21:24:58 :D 21:25:13 no worries... we can revisit later. 21:25:13 swing and a miss? 21:25:16 :P 21:25:19 lol 21:25:27 rayvd: any news on the fuse stuff? 21:26:04 nirik: should have an email sent out to everyone this weekend. 21:26:11 worked out the fu to get the list of all the packages and owners. 21:26:17 thanks to mailing list 21:26:21 cool. Excellent. Yeah, I saw that on the list... 21:27:12 ok, moving on then. 21:27:17 #topic RHEL 5.4 21:27:27 mmmmm RHEL5.4 21:27:35 smooge wanted to mention to folks to help out with CentOS'es QA efforts. 21:27:44 I'm in 21:27:54 I don't know how to contact them about helping, but I suspect the tester list or devel list there would know. 21:28:14 you could ask in #centos-devel 21:28:14 I tried to "apply" as a helper a while back on the centos-devel list and was basically told "support people in irc and write crap on the wiki, now go the hell away" 21:28:36 yeah, becoming involved in centos takes a lot of persistence :) 21:28:43 they want you to fix bugs on the bugs site first. 21:29:01 if they have bugs, doesn't that mean upstream has bugs? 21:29:11 I mean, fixing that would require RH to accept the patches we turn in :) 21:29:30 sometimes they have bugs unique to centos though... 21:29:31 in any case, if anyone reading wants to help contact the centos folks for details. ;) 21:29:32 but yeah :) 21:29:56 Is there anything else with 5.4 we need to do? 21:29:56 stahnma: more or less, yeah ... but I've heard rumors of the CentOS guys sending patches up to redhat 21:30:08 maxamillion: did you ever find out if iotop can come into epel now? 21:30:18 nirik: it would require a port to python 2.4 21:30:27 nirik: it currently requires python 2.5 or newer 21:30:31 ah, I thought it was just a kernel support issue... 21:30:38 it was that too 21:30:41 and they added the needed counters. 21:30:43 but now the python problem has come to light 21:31:07 ah well. Does anyone seem willing to do that porting? 21:31:20 nirik: I plan to look at it 21:31:28 I'd really like to have it in EPEL 21:31:38 yeah, it's handy. 21:32:15 I plan to ping the upstream developer and see what exactly is being used that doesn't have a python 2.4 equivalent/work-around 21:32:22 actually ... lemme do that righ tnow 21:32:24 right now* 21:32:50 ok, if no other 5.4 business we can move on I guess. 21:32:55 maxamillion: thanks. 21:33:07 maxamillion: time permitting I can help look into that as well 21:33:34 #topic newer packages brainstorming 21:33:47 so, mmcgrath brought up the idea of doing some newer packages repo on the list. 21:33:49 derks: sweet! help is more than welcome, free time is hard to come by :/ 21:34:17 Is that something we want to discuss? There was another repo doing this that announced on list... 21:34:33 nirik: that's me... IUS Community 21:34:38 so, this might be just something epel isn't interested in doing. How do folks feel about the basic idea? 21:34:49 derks: ah ha. ;) 21:34:51 http://iuscommunity.org 21:35:06 derks: oh man, that's right! you're the man, I didn't know about that and now I've already got it in use :) 21:35:24 we struggle for volunteers/contributors now. I honestly don't know if we could sustain another repo. 21:35:27 I started the IUS project due to the demand that our customers have for the latest and greatest versions of only a small set of packages: PHP/MySQL/python.. etc 21:35:36 I think we should contribute to IUS in the event someone wants something of that sort 21:36:10 I know its not open to new developers, but people could file tickets and such in the lp.net instance 21:36:12 yeah, one thing I like about epel is that it's very clear and easy to understand that we never replace rhel packages... doing something like this would confuse that message. 21:36:31 nirik: completely agreed 21:36:38 and there isn't likely any way to do parallel installable all the time for things people need. 21:36:56 I agree.. EPEL should not be replacing RHEL packages.. 21:36:58 derks: what packages do you have currently? also, note that there is centos plus for some things. 21:37:02 there are a few things I clobber in my rhel instances, but I like that it's my choice, not a third party repo's. 21:37:26 php52, php53, mysql50, mysql51, python26 (plus a few supporting python26 packages) 21:37:32 all following upstream stable 21:37:41 postgresql84 is in the works 21:37:47 and those are parallel installable? or conflict? 21:38:20 nirik: well.. both. everything conflicts except python26 21:38:28 it was a hot debate on how best to do it 21:38:50 but i've found that if you are wanting PHP 5.2 at the latest... you want everything to work as if you upgraded PHP 21:38:54 and not parrallel 21:39:15 right. conflicts also can prevent the two from confusing where a bug is. 21:39:21 right 21:39:32 personally i think it needs to be decided on a per package basis 21:39:38 mmcgrath: you still around? want to chime in from your point of view? 21:39:47 python obviously has to be parallel... but mysql? 21:40:04 nirik: I'm really just in the boat where I want the packages and wanted to hear what others thought about it. 21:40:12 I'm a bit conflicted as to whether or not EPEl should get into it or not 21:40:33 Though I agree if we did do it, it would probably conflict with EPELs stated goals 21:40:43 I'm completely open to discussion on the direction of IUS 21:40:54 * nirik also thinks quaid's post on the list was right... it would need a new brand/name. 21:41:05 why postgresql84 when https://projects.commandprompt.com/public/pgcore ? 21:41:36 :inode0: one repo, one source 21:41:55 inode0: one person or group to yell at when it breaks 21:42:16 derks: are you open to making the repo more community/open? ie, would you be interested in making a new fedora related project like EPEL for this? 21:42:27 that and IUS generally tries to follow fedora/rhel packaging standards... making for smooth transition 21:42:43 nirik: no doubt 21:43:03 only reason the ius project isn't 'open' to developers now is because I wanted to get everything ironed out 21:43:13 and set a standard for the project 21:43:24 derks: fair enough. 21:43:58 one nice thing about epel is that we can say 'you must meet all the fedora guidelines' and we have all the nice Fedora Infrastructure/accounts/builders/etc... (thanks mmcgrath ! :) 21:44:05 just sent the email to the iotop upstream developer :) 21:44:07 we've been maintaining these packages internally (rackspace) for 3 years now... can't just drop it out there without a bit of care 21:44:37 derks: sure, understandable. 21:45:06 nirik: I think epel is great. and mmgrath I'd be open to discussing a fedora related project like epel 21:45:39 I'd participate, we've been maintaining our own php5.x on RHEL4 for a couple years now 21:45:48 derks: yeah, might be worth thinking about... it could share a lot of the epel setup, but have a different brand/name... 21:45:57 might as well allow others to enjoy the benefits 21:45:58 just brainstorming. ;) 21:46:12 nirik: that is another reason that packaging is closed... we have an internal build system 21:46:31 and offloading that to a process that is already setup would be ideal 21:46:37 * quaid comes in late and reads buffer 21:46:49 *still shakes head at the koji setup* 21:46:52 quaid: we all decided you'd buy us dinner at the next FUDCon :) 21:47:10 * stahnma hums some Rush tunes 21:47:13 derks: yeah... our setup works. ;) 21:47:32 anyhow, I guess we can continue to discuss on list/irc and see what we can make of things... 21:47:41 nirik: I'm still yet to take on any packaging duties as I haven't quite gotten to spend enough time learning it all 21:47:59 that said, our internal build system is built around mock and very similar to koji 21:48:07 thats good. 21:48:38 anything else on newer packages topic? 21:48:50 * nirik should perhaps wait for quaid to read backscroll. 21:48:52 i'm interested to see what you'r alls ideas are for maintaining newer packages 21:49:08 +1 to us looking at Fedora collaborating on a different, compatible EL packaging solution. 21:49:09 derks: koji is pretty easy to setup 21:49:18 are newer packages for EL and Fedora Legacy or LTS roughly the same ? 21:49:40 stahnma: pieces of a related puzzle? 21:49:54 well, either way, I get newer stuff, more often...I think 21:50:00 nirik: thanks, done reading 21:50:16 dgilmore: good, we're going to have to setup our own build system here at work ... more and more custom packages are being requested 21:50:24 :/ 21:50:38 I would say we should even see if we can contribute Fedora Infrastructure bits, but have a third-party branded hosting experience. 21:50:46 maxamillion: just get a form email that says 'no'. ;) 21:50:54 i.e., do what we say we can do -- recreated, derive from the Fedora + EPEL tools and processes 21:51:11 for seeding a new project, with Fedora as an organizational collaborator. 21:51:20 21:52:56 well, If we had enough people interested I could see us adding another project not named EPEL that does this. I'm not sure we have enough folks interested, so I would say for now lets point at derks repo and see if we can gather enough to make doing something in FI worthwhile. 21:53:35 I think there are a lot of consumers interested, but not sure how that translates into maintainers maintaining. ;) 21:53:35 nirik: I'm totally open to the idea of possibly eventually merging IUS over to a collaborative effort 21:53:55 derks: cool. How much interest have you had so far? lots? 21:54:10 nirik: the one thing i think has to be set is... not maintaining newer versions of everything... keep it small 21:54:32 derks +1 to keep it small :) 21:54:41 nirik: $boss seems to say otherwise 21:54:44 yeah, thats hard to do sometimes too... some random person may really want some niche thing. 21:54:45 well.. within the last week just a few subscribers. but like i said we've been doing this internally at rackspace for our customers for years 21:55:00 thousands of customers using latest PHP and MySQL 21:55:30 latest as in php-5.2.10, and mysql-5.0.84 (soon to be 85) 21:55:34 perhaps a 'only stable upstream releases' would cut it down some... depending on the package. 21:56:01 nirik: that's the direction of IUS... 21:56:16 anyhow, I think we can discuss more on this on list/in #epel and see what we can do. ;) 21:56:28 upstream releases plus fixes when possible 21:56:30 * nirik is going to move on to open floor unless someone has something else for this. 21:56:49 #topic Open Floor 21:56:55 anyone have anything for open floor? 21:57:03 neg 21:57:13 dgilmore: I have a quick question if you are still around... did the builders get updated for 4.8 ? 21:57:31 nirik: happens automatically 21:57:54 cool. I didn't know if it did as infrastructure doesn't have any 4.x machines that I know of 21:58:00 nirik: the job the mirrors rhn is tagged to the regullar release 21:58:07 no zstream anything 21:58:14 ok, good to know. 21:58:15 so when rhel bumps we bunp 21:58:47 nirik: there is not 4.x boxes in infra anymore 21:59:00 ok, anything else for open floor from anyone? or shall we close out the meeting? 21:59:04 gotta run 21:59:09 not for me 21:59:18 thanks for coming everyone. 21:59:30 #endmeeting