Meeting Log - 2008-02-21
ricky at fedoraproject.org
Thu Feb 21 20:49:12 UTC 2008
15:00 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure -- Meeting Who's here?
15:00 < mmcgrath> Who's here?
15:00 * iWolf is here
15:00 < ivazquez> Pong.
15:00 -!- petreu [n=peter at fedora/Standby] has quit Read error: 104 (Connection reset by peer)
15:00 * dgilmore
15:00 < dgilmore> skvidal: that was for you
15:00 < skvidal> what
15:00 < skvidal> oh
15:01 * nirik is lurking in the spectator bleachers.
15:01 < skvidal> sorry
15:01 < skvidal> doing other stuff
15:01 < abadger1999> here
15:01 -!- caillon [n=caillon at 126.96.36.199] has left #fedora-meeting ["Leaving"]
15:02 * ricky
15:02 < yingbull> ping
15:02 < yingbull> pong, whichever
15:02 -!- notting [n=notting at redhat/notting] has joined #fedora-meeting
15:02 -!- MrBawb [i=abob at guppy.drown.org] has joined #fedora-meeting
15:02 < ricky> Bad time for me to start eating a mango :(
15:03 < jcollie> ping
15:03 < mmcgrath> heh no worries.
15:03 * f13
15:03 < mmcgrath> alrighty, lets get started
15:03 -!- petreu| is now known as petreu
15:03 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure -- Tickets
15:03 < mmcgrath> .tiny https://fedorahosted.org/fedora-infrastructure/query?status=new&status=assigned&status=reopened&group=milestone&keywords=%7EMeeting&order=priority
15:03 < zodbot> mmcgrath: http://tinyurl.com/2hyyz6
15:04 < mmcgrath> so first ticket
15:04 < mmcgrath> .ticket 347
15:04 < zodbot> mmcgrath: #347 (Set localtime on all our servers to UTC) - Fedora Infrastructure - Trac - https://fedorahosted.org/projects/fedora-infrastructure/ticket/347
15:04 -!- daMaestro [n=jon at fedora/damaestro] has joined #fedora-meeting
15:04 < mmcgrath> dgilmore: any word on that?
15:05 -!- wolfy [n=lonewolf at fedora/wolfy] has left #fedora-meeting ["If you don't know what procrastination is just look up the definition tomorrow."]
15:05 < dgilmore> mmcgrath: I was thinking we should be able to use puppet to push out the timezone files to each box
15:06 < dgilmore> and then reboot everything
15:06 < mmcgrath> <nod> that works for me, right now its not puppet managed but I don't see any reason why it shouldn't be.
15:06 < jcollie> i do that on my boxes at work, although not with puppet
15:06 < dgilmore> i think using puppet will make it much simpler to do that lagging in and running commands on each and every box
15:06 < mmcgrath> dgilmore: I remember you saying you'd be doing it during a trip or something, when was that scheduled for?
15:06 < dgilmore> mmcgrath: ill be in boston all next week
15:07 < mmcgrath> Should we schedule some outage for it?
15:07 < dgilmore> so i figured id do it about midnight wednesday or thursday
15:07 < dgilmore> yeah we should
15:07 < dgilmore> lets schedule it for thursday
15:08 < mmcgrath> sounds good to me, mind sending out an outage notification? I'll make sure I'm around then as well, shouldn't take too long.
15:09 < f13> thursday works for me.
15:10 < mmcgrath> k, then we'll get that all setup..
15:10 < mmcgrath> anyone have any concerns about that before moving on?
15:10 < jcollie> not from me
15:10 < dgilmore> ill send out the notice
15:11 < mmcgrath> thanks
15:11 < mmcgrath> alrighty, next ticket.
15:11 < mmcgrath> .ticket 365
15:11 < zodbot> mmcgrath: #365 (mailman setup (project/hosted)) - Fedora Infrastructure - Trac - https://fedorahosted.org/projects/fedora-infrastructure/ticket/365
15:11 < mmcgrath> jcollie: want to take the floor on that one?
15:11 < jcollie> sure
15:11 < jcollie> i think we're ready to go, except for documenting a SOP
15:12 < jcollie> which would be SOOO much easier if we were on mediawiki but that's another topic
15:12 < mmcgrath> heheheh
15:12 < mmcgrath> cool
15:13 < mmcgrath> jcollie: once the sop's done just close that ticket and you have a huge thankyou from us and a bunch of other fedorahosted users, that will be most helpful to finally deploy.
15:13 < jcollie> there are a couple of lists out there so it wouldn't hurt for people to subscribe and start sending some mail through
15:13 < mmcgrath> alrighty, next ticket
15:13 < mmcgrath> .ticket 270
15:13 < zodbot> mmcgrath: #270 (Fedora Wiki allows editing raw HTML) - Fedora Infrastructure - Trac - https://fedorahosted.org/projects/fedora-infrastructure/ticket/270
15:13 -!- viking-ice [n=johannbg at dsl-149-97-225.hive.is] has joined #fedora-meeting
15:13 < mmcgrath> ricky: paulobanon: I'm not sure what the latest is on that to be honest.
15:13 -!- jnettlet__ [n=jnettlet at c-76-118-159-90.hsd1.ma.comcast.net] has joined #fedora-meeting
15:13 -!- viking-ice [n=johannbg at dsl-149-97-225.hive.is] has quit Connection reset by peer
15:13 < mmcgrath> Maybe its something we should't put much work into until after the wiki migration?
15:14 < ricky> Sorry for letting that sit - hopefully, it won't be a problem anymore soon :)
15:14 < jcollie> how fast do we think we can move to mediawiki?
15:14 < ricky> We have a pretty nice-looking setup already, thanks to mizmo.
15:14 < mmcgrath> jcollie: well, I've got to go around to the teams and make sure everyone's ok.
15:14 < mmcgrath> here's the hangups that I still know about
15:14 < ricky> I assume that we want FAS integration from the get-to?
15:14 < ricky> **get-go
15:14 < mmcgrath> 1) docbook (this is a must)
15:14 < mmcgrath> 2) FAS integration
15:14 < mmcgrath> 3) Make sure the art attachments migrated properly.
15:15 -!- fcrippa [n=fcrippa at host75-233-static.28-87-b.business.telecomitalia.it] has joined #fedora-meeting
15:15 < mdomsch> howdy
15:15 < mmcgrath> mdomsch: howdy.
15:15 < mmcgrath> so those are the big hanups.
15:15 < mmcgrath> ricky: mind taking the Meeting keyword out of that?
15:16 < ricky> mmcgrath: Sure thing.
15:16 < mmcgrath> alrighty
15:16 < mmcgrath> next ticket then.
15:16 < mmcgrath> .ticket 395
15:16 < zodbot> mmcgrath: #395 (Audio Streaming of Fedora Board Conference Calls) - Fedora Infrastructure - Trac - https://fedorahosted.org/projects/fedora-infrastructure/ticket/395
15:16 < mmcgrath> I don't think there's been any new progress on that but thought I'd check.
15:16 < jcollie> not much has happened lately...
15:16 < mmcgrath> jcollie: anything?
15:16 < mmcgrath> no worries, we've got time to work on that one.
15:17 < jcollie> but we've got 1+ month to get that into shape
15:17 < ivazquez> Do we have Asterisk/Callweaver/w/e up yet?
15:17 < mmcgrath> Ok, so thats all the tickets. I wanted to switch over to something that some of us have been working pretty hard at this week.
15:17 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure -- Buildsystem.
15:17 < f13> que the dramatic lights
15:18 < mmcgrath> ivazquez: yeah, Asterisk is up but not fully deployed (no fas integration, no mailboxes, etc) but we have had meetings on it.
15:18 < jcollie> ivazquez: yes -- asterisk1 is up and running but it's still in a beta state
15:18 < mmcgrath> ok. so the buildsystem
15:18 < mmcgrath> for those of you that haven't seen the nagios alerts we've had 8 outages of the buildsystem since monday.
15:18 < dgilmore> :(
15:18 < mmcgrath> coming at all hours of the day or night.
15:18 < mmcgrath> Its not been fun.
15:18 < mmcgrath> so whats been done?
15:18 < jcollie> yeah that blows major chunks
15:19 < mmcgrath> well, we've spent some time getting the database in better shape.
15:19 < mmcgrath> we've made the website not hit the rpmfiles table which made a huge difference in general load on the db, but didn't help the spikes.
15:19 < mmcgrath> we added a robots.txt file, banned the soc search engine from crawling it.
15:20 < jcollie> do we want to ban all search engines?
15:20 < f13> and we threw 2K tasks at it
15:20 -!- ChitleshGoorah [n=chitlesh at 188.8.131.52] has joined #fedora-meeting
15:20 < dgilmore> we should add those indexes to the task table
15:20 < mmcgrath> jcollie: until I can sleep through the night we do.
15:20 -!- knurd is now known as knurd_afk
15:20 < mmcgrath> <nod>
15:20 < mmcgrath> there's an index we got rid of and an index we're looking to add.
15:20 < mmcgrath> .taskload
15:20 < zodbot> mmcgrath: Tasks running - Open: 38 Total: 253
15:20 < abadger1999> (two indexes to add)
15:20 < f13> it's been moving very fast today
15:20 < f13> as far as chewing through open tasks
15:21 < mmcgrath> dgilmore's also working on moving the hub away from the website.
15:21 < mmcgrath> /koji/ is the website and /kojihub is the xmlrpc interface.
15:21 < f13> new repo tasks are faster than ever.
15:21 < f13> close to 10 minutes now
15:21 < f13> http://koji.fedoraproject.org/koji/taskinfo?taskID=457867
15:22 < dgilmore> f13: still to long i think but its better
15:22 < mmcgrath> interestingly we've seen many times where the hub is nice, peppy and responsive and the website isn't, but since they're both on the same box right now, there's no way to stop or restart one without the other.
15:22 < f13> dgilmore: it's only 4~ minutes for the query before it hands it off to the builder to createrepo
15:23 < f13> dgilmore: the fact that we could get a 13 minute newRepo task done, while the builders have 300~ queued tasks is pretty amazing in my book
15:23 < mmcgrath> I think there's still a lot of work todo but with the exception of some timeouts this morning (very short lived) from nagios to check swap, I've not seen any actual outages yet.
15:23 < f13> yeah, today has been quite nice
15:23 < f13> I think splitting the hub from teh website is a good step
15:23 < mmcgrath> f13: how long did you say it takes to do the whole signing process?
15:23 < f13> and I'd like to take that further by having the repos stuff done by something else too
15:24 < mmcgrath> and is that something that blocks on disk IO or koji io or both?
15:24 < f13> mmcgrath: signing of packages depends on how many packages, but can often take a couple hours
15:24 < f13> mmcgrath: both.
15:24 < dgilmore> f13: im seeing about 17 minutes on sparc http://sparc.koji.fedoraproject.org/koji/taskinfo?taskID=50738
15:24 < f13> signing does in this order:
15:24 < f13> query koji for existing cached or written out signatures for a set of biulds.
15:24 < f13> copy unsigned rpms to a work directory
15:24 < f13> rpm sign the rpms
15:24 < f13> import the signed headers into koji
15:25 < f13> ask koji to write out copies of the packages with the signed headers to the koji directories.
15:25 < dgilmore> f13: can you approve my announcement please :)
15:25 < skvidal> f13: I looked at this step in koji
15:25 < skvidal> [15:20] <f13> rpm sign the rpms
15:25 < f13> the query and the import take a long time, but then again so does the copying/signing if there ar ea lot of builds.
15:25 < skvidal> does it just call rpm --resign with xargs?
15:25 < f13> skvidal: 'in koji' ?
15:25 < f13> the signing doesn't happen in koji
15:25 < skvidal> well, I looked to see where it occurred
15:25 < mmcgrath> f13: Also I put in a bug to koji to hash the tasks directory, hopefully that will help us some.
15:25 < skvidal> and it doesn't seem to be called anywhere
15:25 < skvidal> ok
15:26 < skvidal> that's what I thought was going on
15:26 < skvidal> so i'm not losing my mind
15:26 < f13> skvidal: signed_unsigned is a terrible terrible script in the releng git repo
15:26 < skvidal> okay, I kinda got that feeling
15:26 < skvidal> thanks
15:26 < f13> it's slated to be replaced by sigul, although some parts of it will survive as a client of siguls to query koji for what neesd to be signed.
15:27 < f13> sign_unsigned came from inside RH
15:27 * skvidal is shocked to hear that
15:27 < mmcgrath> heh
15:28 < mmcgrath> alrighty so thats really whats been going on there.
15:28 < mmcgrath> its been sucking up a lot of our time this week but hopefully some of our problems can be bandaided until the permanent fixes can be put in. Koji's been deployed for almost a year now, there's still a long way to do.
15:28 < mmcgrath> koji is a journey, not a destination :)
15:29 -!- kital [n=Joerg_Si at fedora/kital] has joined #fedora-meeting
15:29 < mmcgrath> alrighty, so thats really all I have that I really wanted to talk about this meeting so lets open the floor.
15:29 < ricky> Wow, only one year.
15:29 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure -- Open Floor
15:29 < smooge> just as long as cooker isnt the destination
15:29 -!- jnettlet_ [n=jnettlet at c-76-118-159-90.hsd1.ma.comcast.net] has quit Connection timed out
15:29 < mmcgrath> Anyone have anything they'd like to discuss?
15:30 < jcollie> i've been working on a tg based openid server for a day or so
15:30 < ricky> jcollie: Oooh, that'd be really handy in FAS!
15:30 < mmcgrath> jcollie: how's that going? It'll be nice to have some more openid expertise around here.
15:30 < jcollie> so it should be pretty easy to use your fedora id to log into stuff
15:30 < ricky> The provider that I have there is hideous (and probably only a partial/incorrect implementation of the spec)
15:30 < jcollie> i've had a couple of successful logins with it but somethings broken right now
15:31 < jcollie> i'm using the python-openid libs so we should be good as far as standards go
15:31 < daMaestro> +1 on openid
15:31 -!- spoleeba [n=one at fedora/Jef] has quit "Leaving"
15:31 < daMaestro> i'd love for FAS users to be able to login to the unity stuff
15:31 < lmacken> awesome.
15:32 < lmacken> jcollie: did you get the nested arg stuff working ?
15:32 < ricky> I'm curious as to how you're testing it.
15:32 < jcollie> lmacken: yes, thanks for the help
15:32 < GeroldKa> ping kanarip
15:32 < lmacken> great. no problem
15:32 < kanarip> GeroldKa, pong
15:32 < abadger1999> jcollie: What are your plans for that? Separate server? Integrate into FAS2 if we're interested (We are!)
15:32 < ricky> I think I used http://openidenabled.com/resources/openid-test/diagnose-server, but it might not be the most up-to-date/complete.
15:33 < daMaestro> i do have a question about my upcoming attempt to migrate everything from moinmoin to zwiki; am i just wasting me time? would it at least be interested to see how zope handles our current data set?
15:33 < daMaestro> s/interested/interesting/
15:33 < jcollie> abadger1999: it used the tg identity stuff so we should be able to plug in a fas identity provider
15:33 < jcollie> er uses
15:33 < ricky> jcollie: Ohh, so it's a client, not a provider?
15:33 < jcollie> no it's a provider
15:33 < ricky> Or it could be both.. hmm.
15:34 < jcollie> we would need to write a separate consumer
15:35 < jcollie> code is here: http://jcollie.fedorapeople.org/samadhi.git/
15:35 < jcollie> i'll probably request hosted space soon
15:35 < ricky> abadger1999: Do you know if we can have multiple auth providers in a TG project?
15:35 < lmacken> jcollie: great name :)
15:35 < jcollie> lmacken: thought you'd like it
15:36 < abadger1999> ricky: I think the web app only knows about a single provider.
15:36 < jcollie> dl the code and look at the logo i ginned up quick
15:36 < abadger1999> That provider could get information from multiple sources, though.
15:36 < ricky> Ah, great.
15:36 < jcollie> eventually i'll need some help prettifying the html... i'm no good a css and all that
15:37 -!- fabian_a [n=fabian at 84-75-160-95.dclient.hispeed.ch] has joined #fedora-meeting
15:37 < mmcgrath> that will be great to finally get installed.
15:37 < ricky> jcollie: Anyway, when you get a chance, we'd love you to take a look at OpenID in FAS :)
15:37 < mmcgrath> Alrighty, anyone else have anything they'd like to discuss?
15:38 < ricky> I'd be glad to nuke whatever hideousness I put there before.
15:38 < daMaestro> mmcgrath, is my attempts with zwiki futile?
15:39 < mmcgrath> daMaestro: probably
15:39 < daMaestro> ok
15:39 < mmcgrath> unless you want the experience of doing it :)
15:40 < mmcgrath> alrighty, anyone have anything else? If not I'll close the meeting.
15:40 < dgilmore> nothing else?
15:40 < dgilmore> :)
15:41 < mmcgrath> Alrighty!
15:41 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure -- Meeting closed
15:41 < mmcgrath> thanks for coming everyone!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 189 bytes
Desc: not available
More information about the Fedora-infrastructure-list