fedora-cloud
LOGS
20:59:44 <gregdek> #startmeeting
20:59:45 <zodbot> Meeting started Thu Mar  4 20:59:44 2010 UTC.  The chair is gregdek. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:59:46 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
20:59:51 <ewan> I had an action to look at Ubuntu's init scripts; have been doing that, and am in the process of beating them into an RPM.
21:00:05 <gregdek> ewan... nice!
21:00:12 <gregdek> #topic Roll Call
21:00:19 <gregdek> Roll call!  Who's around?
21:00:29 * gholms sips his tea
21:00:31 <mmcgrath> I'm here
21:00:43 <smooge> here
21:00:52 <gregdek> jforbes gholms?
21:01:00 <gholms> gregdek: Hi!
21:01:03 <gregdek> Ah, gholms sipped his tea.
21:01:05 <gregdek> I missed it.  :)
21:01:14 * jforbes is here
21:01:30 <gregdek> That's everyone we need most, I think...
21:01:33 <gholms> Where should we start?
21:01:41 <gholms> ewan, maybe?
21:01:54 <gregdek> #topic Ewan EC2 scripts
21:02:03 <gregdek> I will parrot:
21:02:06 <gregdek> <ewan> I had an action to look at Ubuntu's init scripts; have been doing that, and am in the process of beating them into an RPM.
21:02:11 <gregdek> So that's good stuff.
21:02:15 <gregdek> ewan, any eta?
21:02:45 <gregdek> Or did we already lose him?  :)
21:02:49 <gholms> What is there of interest in Ubuntu's init scripts, out of curiosity?
21:03:07 <gregdek> Good question.
21:03:21 <gregdek> But since ewan appears to have already gone into lurk mode...
21:03:33 <gregdek> ...I propose we just move on for now and pursue on-list?
21:03:35 <gholms> Oh, hang on a sec
21:03:51 * gregdek hanging
21:03:51 <gholms> #info ewan working on packaging Ubuntu's EC2 init scripts
21:03:58 <gholms> Ok, I'm done.  :)
21:04:01 <gregdek> :)
21:04:16 <gregdek> #topic jforbes: kernel wranglement
21:04:29 <gregdek> jforbes... good news?  :)
21:04:31 <jforbes> Well, we should have some good news tomorrow
21:04:37 <jforbes> 2.6.32 has enough karma
21:04:53 <jforbes> But it hasn't been moved from testing to stable, I expect that will happen tonight
21:05:06 <gregdek> #info 2.6.32 should move into F12 stable by 3/5
21:05:16 <gholms> Sweet
21:05:28 <gregdek> And then we can start talking about uploading AKIs.
21:05:34 <jforbes> Right
21:05:44 <gholms> Did we ever figure out how we plan on building initramfs images?
21:05:45 <gregdek> Except that "things about Amazon are about to get easier in that regard" in a way that we can't discuss yet.
21:06:12 <jforbes> gholms: nice thnig about dracut, it builds an init everything
21:06:31 <gregdek> In other words, nothing to discuss?
21:06:35 <gholms> Sure, but we have to pre-generate them and upload them to S3.
21:06:36 <jforbes> correct
21:06:42 <gholms> What's next?
21:07:15 <smooge> cool what testing do you need?
21:07:18 <gregdek> Gah, I think we already actually closed these...
21:07:20 <gregdek> but...
21:07:33 <gregdek> #topic document kernel/initramfs upload process
21:08:08 * gregdek hrms.
21:08:12 <jforbes> gregdek: same, status as before, kernel is a precursor to documentation since the docs will be confirmed and updated as the AKI/ARI is pushed
21:08:28 <gregdek> Gotcha.
21:08:41 <gregdek> Moving on:
21:08:57 <gregdek> #topic share upload credentials with Fedora kernel developers
21:08:59 <gregdek> The same?
21:09:50 <jforbes> yup, need the docs before we share the credentials
21:09:58 <gregdek> OK.
21:10:13 <gregdek> Moving on:
21:10:19 <gholms> #info need docs before sharing credentials
21:11:11 <gregdek> #topic mirroring various things (srpms, mirror manager, etc.) at Amazon
21:11:35 * mdomsch perks up
21:11:39 <gregdek> mdomsch, did you get the latest email exchange?
21:11:45 <gregdek> To update others:
21:12:03 <mdomsch> nathan bounced the questino of how RHEL is mirroed to Mike Ferris
21:12:09 * mdomsch can't type
21:12:14 <gregdek> Working with Nathan Thomas at Amazon, clarifying what will be required to get MM up and running for Fedora.
21:12:32 <gregdek> It appears that RH is also mirroring RHEL, so we're trying to find out if we can piggyback.
21:12:57 <gregdek> mdomsch, I'm leery of waiting too much for responses from the RHEL side of the house...
21:13:09 <gregdek> ...is there any reason not to move forward by just setting up MM machines in EC2?
21:13:26 <gregdek> (Once we actually get F12 images up there?)
21:13:26 <gholms> Who's going to pay for them?
21:13:30 <gregdek> We will.
21:13:33 <gregdek> At first, anyway.
21:13:51 <gholms> Do we know each availability zone's IP ranges?
21:13:57 <gregdek> Transfer rates inside of Amazon will be negligible, so long as we have MM instances in each availability zone.
21:14:06 <gregdek> That's a detail I'm sure we can work out.
21:14:22 <gholms> But we have to get MM to point to the right instance in each zone.  I'm sure we can figure it out.
21:14:31 <gregdek> But yes, it's a question we must ask.  :)
21:14:56 <gregdek> #info Be sure to find out how Amazon maps IP addresses to availability zones, and make sure that MM uses that method
21:15:13 <mdomsch> gregdek - no argument from me
21:15:23 <gholms> We also need to figure out how to do this with elastic IPs.
21:15:24 <mdomsch> gregdek, get the instances, I'll set up the mirrors and MM
21:15:30 <gregdek> mdomsch, done.
21:15:35 <gregdek> Well, it *will* be done.  :)
21:15:59 <gregdek> #action gregdek will get instances for mdomsch for MM asap
21:16:07 <gregdek> (I'll ask for RHEL5 instances.)
21:16:12 <mdomsch> perfect
21:16:47 <gregdek> Moving on:
21:16:48 <smooge> gregdek, are you asking for a full mirror inside of Amazon or a sub mirrror?
21:16:55 <gregdek> Oh, wait.
21:17:02 <gregdek> smooge: it's a good question.
21:17:07 <gregdek> That I don't even understand.  :)
21:17:07 <gholms> We should only mirror releases that are actually on EC2.
21:17:15 <gregdek> Oh. Yes, definitely.
21:17:18 <smooge> because we need 1 TB of disk space per full mirror
21:17:23 <gholms> Otherwise we're just wasting a whole lot of disk space.
21:17:26 <smooge> well 800 GB or so
21:17:37 <gholms> 1 TB runs somewhere around $150 a month.
21:17:52 <smooge> the big eater is updates
21:17:58 <gregdek> That's nothing, honestly.
21:18:06 <gregdek> We've got budget.
21:18:15 <mdomsch> still fair to only mirror what we can use
21:18:16 <smooge> gregdek, more than my budget :)...
21:18:21 <smooge> np
21:18:27 <gholms> That's per mirror, one of which will need to be in each zone.
21:18:31 <gregdek> Yes, i still only want to mirror what's necessary.
21:18:36 * gholms has no idea what our budget is like
21:18:41 <gregdek> But to me, that's an implementation detail.
21:18:44 <mdomsch> and mirror epel too :-)
21:18:47 <smooge> ok I will get an F12 estimate
21:18:55 <smooge> mdomsch, hey you beat me
21:18:57 <gregdek> And our budget is probably going to be about $10k/q for now.
21:19:07 <gholms> Ooh, yes!  EPEL mirrors would be good.
21:19:13 <gregdek> That's pretty much our starting point where we don't even have to ask for more.
21:19:26 <gregdek> More than that is possible, but we'll have to look around for it.
21:19:49 <gregdek> Any more questions around this?
21:19:50 <gholms> Reserved instances reduce costs a bit, though there's the initial charge...
21:20:01 <gregdek> Hm, yes.
21:20:15 <gregdek> Good point: this is probably a perfect use for resevred instances.
21:20:31 <gregdek> #info Consider reserved instances for the MM systems.
21:20:32 <gholms> The only big problem with that is the large lump sum charge every three years.
21:20:40 <gregdek> I'll look into it.
21:20:57 <gregdek> Nice catch, gholms, thanks.
21:21:02 <gregdek> Any other notes/questions?
21:21:13 <gholms> Not for this.
21:21:18 <gregdek> OK, moving on.
21:21:21 <gregdek> #topic euca2ools-1.2 in Fedora updates
21:21:28 <gregdek> They're in!  :)
21:21:41 <gholms> I pushed 1.2 to stable so you can all play with it now.  :)
21:21:54 <gholms> Normally I don't mark updates that don't have any feedback at all as stable, so in the future if there's an update available for testing you should try it and give it +1s/-1s/comments so it doesn't sit in updates-testing forever.
21:22:00 <smooge> what is the difference between eucatools and eucalyptus (osrry for being dumb here)
21:22:13 <gholms> Eucalyptus is the VM server side.
21:22:37 <gholms> e.g., if you want to run your own EC2-ish cluster.
21:22:49 <gregdek> Euca2ools is the client side -- basically a rewrite of the EC2 client tools under an actually open license.
21:22:57 <gholms> euca2ools is the client interface.
21:23:17 <gregdek> To us, it's useful even in the absence of the Eucalyptus server...
21:23:29 <gregdek> ...because it allows any Fedora user to manage EC2 instances with free tools.
21:23:49 <gholms> There are a few operations it doesn't yet support, but I'm sure they'll be added in later releases.
21:23:51 <gregdek> (And one day, when Eucalyptus gets in, they'll be useful for Eucalyptus also.)
21:24:10 <gregdek> smooge, sufficient reply?
21:24:12 <smooge> ok thanks
21:24:29 <gregdek> Any more notes about euca2ools?
21:24:30 <gholms> If you need a function that euca2ools doesn't have, you can find Amazon's tools in RPM Fusion nonfree.
21:24:46 <gholms> Nope.  Just test updates in the future, people.
21:24:53 <gregdek> :)
21:25:08 <gregdek> All right.  We already covered ewan's item.
21:25:17 <gregdek> So that's it for our official list of stuff.
21:25:28 * gholms has a couple things
21:25:32 <gregdek> Shoot.
21:25:45 <gregdek> #topic open floor
21:25:57 <gholms> FESCo wanted us to contact VPS providers that offer old Fedora releases.  Do we need to wait on that?
21:27:13 <gholms> [a cricket chirps in the distance]
21:27:26 <gregdek> Heh.
21:27:35 <smooge> no I don;t think waiting is important
21:27:36 <gregdek> More info, please?
21:27:48 <gregdek> What would we be waiting on, exactly?
21:27:55 * gholms goes to dig out the link
21:28:04 <Southern_Gentlem> !
21:28:19 <gregdek> Southern_Gentlem, yes?
21:28:46 <gholms> http://lists.fedoraproject.org/pipermail/cloud/2010-February/000109.html
21:29:13 <Southern_Gentlem> I talked to acouple of godaddy admins at Southern California Linux Expo an the good news there is that they are gressively moving people off fedora and onto centos
21:29:25 <gholms> *sigh*
21:29:35 <gholms> I wonder if it's CentOS or "CentOS"...
21:30:02 <Southern_Gentlem> eof
21:30:25 <gregdek> Is the goal to say "hey, you guys, stop offering F8, thanks?"  Is that it?
21:30:39 <gholms> "Dear VPS provider, we noticed you provide Fedora 10 and 9, those  releases are end of life. Would you consider offering F12? Would you  like assistance in making an image for your needs? we have a f12 image  in format X,Y,Z available. Please contact cloud-sig@ if you have questions"
21:30:49 <gregdek> Because that will be the obvious next step *after* we make it easy to run F12.
21:31:11 <gregdek> I'm not too worried about it.  I think it's something we will definitely want to do, but nothing blocks one way or the other, I don't thikn.
21:31:13 <gregdek> think.
21:31:28 <gregdek> For Amazon, anyway.
21:31:45 <Southern_Gentlem> gholms,  we still fc2 godaddy boxes still coming into #fedora
21:31:46 <nirik> basically I would like to reduce the number of people who we see in #fedora that have an EOL install and their VPS provider doesn't offer anything newer. ;)
21:31:46 <gholms> We should basically mention that there is a newer release out, and they can come to us with questions.
21:32:04 <gregdek> Yep.
21:32:17 <gholms> We should also watch out for VPS providers that install strange, custom kernels.
21:32:20 <gregdek> When we actually can make that newer release available -- which currently, we can't.
21:32:26 <Southern_Gentlem> unless the provider uses plesk
21:32:57 <gregdek> gholms, good enough?
21:33:05 <gholms> Sounds fine to me.
21:33:25 <gregdek> You had another item?
21:33:38 <gholms> mmcgrath: Did you learn anything interesting about IBM?
21:34:01 <mmcgrath> gholms: not especially, I was trying to rebuild one of their RHEL boxes to Fedora and it proved to be more difficult then I thought.
21:34:13 <mmcgrath> I'm wondering if they have some special bootloader that is preventing me from doing what I was tryin gto do :)
21:34:24 <mmcgrath> I did notice they are firewalling some ports though.
21:34:46 <mmcgrath> oh, and they had some downtime last week because of a storm...
21:34:48 <gregdek> mmcgrath, any policy to not do that?
21:35:05 <gregdek> I know that ec2 allows you to set ports and closes pretty much everything by default.
21:35:22 <mmcgrath> I didn't explicitly see one though all of their offerings were supported (IE: for pay support)
21:35:28 <mmcgrath> which is probably why Fedora wasn't there
21:35:34 <mmcgrath> they might be targeting a different group of people.
21:35:49 <mmcgrath> but still, it's usually not difficult to rebuild a host.  Something on their end is blocking it.
21:36:41 <gregdek> I suspect we'll dig into that whole scene once we're happy with how Fedora+EC2 is working out.
21:36:49 <gholms> Sounds reasonable.
21:37:16 <gholms> Any objections to meeting at this time in #fedora-meeting for future meetings?
21:37:29 <gregdek> I tried today!
21:37:30 <gregdek> :)
21:37:39 <gregdek> I already claimed it in the official schedule.
21:37:52 <gregdek> Today they sort of squatted for the release readiness meeting.
21:38:04 <gregdek> I'll make sure that poelcat doesn't do that next week.
21:38:25 <jforbes> That works for me
21:38:27 <gholms> The alpha isn't supposed to be released twice, so that's a safe bet.
21:38:31 <gregdek> LOL
21:38:36 <mdomsch> gregdek, get F13-Alpha out the door and he won't have to
21:38:40 <gregdek> "Supposed to be" being the key there.
21:38:47 <gholms> #info future meetings to occur at same time in #fedora-meeting
21:38:48 <gholms> :)
21:38:57 <gregdek> mdomsch, yeah, because that's all me.  LOL
21:39:01 <smooge> he could have used fedora-meeting-1
21:39:12 <gregdek> I can't even spell F13-Alpha.
21:39:16 <gholms> Apparently there's a #fedora-meeting-2 as well.
21:39:56 <gholms> Anyone have anything else for today?
21:40:04 <gregdek> Yeah, but if we were forced to do that, I'd just as soon squat here.  The idea is to get more attendance, and I can't imagine meeting-2 would help there.
21:40:09 <gregdek> No, that's all I have.  Anyone else?
21:40:13 <jforbes> nothing here
21:40:20 <gholms> Nah, *they* should've moved.
21:40:44 <gregdek> :)
21:40:46 <gregdek> OK, folks.
21:40:59 <gregdek> Then we're done in 30 seconds if no objection.
21:41:15 <gholms> OBJECTION!
21:41:20 <gholms> :P
21:41:32 <gholms> Anyone here working on Eucalyptus RPMs?
21:42:26 <gholms> I guess not.
21:42:33 * gholms wanders off to make more tea
21:42:41 <gregdek> gholms, I saw some talk on the list...
21:42:53 <gholms> Yeah, did anything come of that?
21:43:43 <gregdek> I didn't see any followup.  Poke the list, I'd say.
21:43:53 <gregdek> And I may poke a few people off-list.
21:43:58 <gregdek> Anyway...
21:44:00 <gregdek> #endmeeting