20:01:54 <decause> #startmeeting
20:01:54 <zodbot> Meeting started Wed Aug 12 20:01:54 2015 UTC.  The chair is decause. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:01:54 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
20:02:25 <decause> #topic What Does Red Hat Want? A Semin-Philosophical Look at the Fedora/Red Hat relationship
20:02:42 <decause> #info Denise Dumas, VP of Platform Engineering (RHEL)
20:03:05 <decause> I'm responsible for hte teams that run RHEL, and the teams that provide support for Fedora Infrastructure, which is actually a great place to be
20:03:15 <decause> this puts me into a great place to talk about "What does Red Hat Want?"
20:03:24 <decause> first, foremost, Three is no "one voice" of Red Hat
20:03:41 <decause> A group of us sat down to figure out what message to bring to a Fedora office
20:03:50 <decause> some of you do, but most of you don't work at Red Hat
20:04:04 <decause> so, talking about where RHT thinks the world is going, and what working with Fedora looks like would be good
20:04:12 <decause> #topic Red Hat the Company
20:04:19 <decause> is about providing support to enterprises
20:04:34 <decause> genreally, the folks that work on Fedora also work on enterprise software in their dayjobs
20:05:05 <decause> Our RHEL Customers value stability, yet demand innovation.
20:05:13 <decause> "It must be a floorwax, and a desert topping"
20:05:20 <decause> it must be stable, and it must be bleeding edge
20:05:27 <decause> We try very hard at RHT, to cope with this
20:05:37 <decause> this describes well the place we're trying to get to
20:05:46 <decause> stable enough, moving fast, it can't break
20:05:51 <decause> this is where we want RHEL to cover
20:05:57 <decause> Fedora, has a diff set of Criteria
20:06:00 <decause> #topic Fedora
20:06:03 <decause> is about innovatoin
20:06:15 <decause> from a RHT perspective, innovation is only something we want to pull in when it is stable enough
20:06:29 <decause> Fedora is where you get to have fun, and experiment, and see what looks like or turns out to be the right combinatoin
20:06:32 <decause> it is fast moving
20:06:39 <decause> my god it is fast moving in a 6 month cycle
20:06:48 <decause> pulling that off with RHEL would be a miracle
20:06:52 <decause> Fedora is also about community
20:06:59 <decause> RHT recognizes we're not in this alone
20:07:14 <decause> innovation happens so quickly, in upstreams, it is not something any one company can control
20:07:27 <decause> any proprietary company that thinks like that is going to be roadkill
20:07:42 <decause> community support is what we want to provide for Fedora
20:07:48 <decause> and we do a lot in Fedora
20:07:57 <decause> Fedora is a chance to try out htings, see what resonates
20:08:05 <decause> there are 15 different packages for any given usecase, right?
20:08:24 <decause> some are going to be winners, some are not, some with a little assistance, and some patches upstream, are going ot have a much higher chance of viablility.
20:08:47 <decause> we want to see, in Fedora, how things play out by kicking the tires, and finding the difficulties, and find ideas with traction with the people who may someday pay for support.
20:09:03 <decause> Fedora as you probably know, becomes the basis for RHEL
20:09:26 <decause> we watch Fedora release, releasa, reelase, and after 4-5-6-7 of them, we'll bring one in house, and start building it in house, and see which packages need/don't need to be there
20:09:30 <decause> understand deps
20:09:32 <decause> work on hardening
20:09:36 <decause> work on security
20:09:42 <decause> then make a major release for RHEL
20:09:45 <decause> it is a bottom up process
20:09:48 <decause> there are lots of upstreams
20:09:54 <decause> they move at their own pace
20:09:59 <decause> the kernel is off at their own pace
20:10:06 <decause> systemd moderately takes input, if we're lucky
20:10:29 <decause> we try to pull it all together and make into a cohesive whole, and create a marketing story that says "look, we meant to do that!"
20:10:34 <decause> that has been much of the history
20:10:46 <decause> if you think back a few years to Fedora Rings proposal, and some of th ework to support Fedora Editions
20:11:04 <decause> some of that has been RHT the ocmapny, working with RHT the smart people who work for us, to find better ways to support a distro
20:11:21 <decause> we're trying to be a little more "planful" about the way that we, RHT, approach an OS product
20:11:29 <decause> when you think about product, there is bottom-up, and top-down
20:11:34 <decause> we'll never be top-down at RHT
20:11:42 <decause> product management has to be the toughest job in the world
20:11:54 <decause> we'll never be standing up dictating "this is what we do, now do it!"
20:12:03 <decause> all our engineers would quit, first of all
20:12:08 <decause> our engineers are *really* smart
20:12:16 <decause> we need to understand, from them, what is going on in FOSS world
20:12:20 <decause> tha tinforms what goes into the product
20:12:34 <decause> we're trying to be a little more thoughtful about the guidance we give to an engineering team
20:12:39 <decause> they make lots of decisions, everyday
20:13:03 <decause> they make them indepependently, becuase no one can micromanage an engineer... there are not enough management cylces in the world, an that is a stupid waste of cycles too
20:13:18 <decause> RHT wants to give our teams a general idea of where the product we'd like to go, and a genreal direction
20:13:31 <decause> there will be lots of decisions along the way, but we want to get to this destination
20:13:36 <decause> when you think abou tRHEL, think abou tthat
20:13:45 <decause> setting general direction for engineering
20:13:51 <decause> that will always be something that starts in FEdora
20:13:59 <decause> it flows into RHEL from there
20:14:05 <decause> RHT, the company has a mission statement
20:14:08 <decause> #topic RHT Mission
20:14:35 <decause> "We want to be the catalyst in communities of customers, contributors, and partners creating better technology the open source way."
20:14:42 <decause> you may remember the posts to memo list
20:14:50 <decause> mizmo I'm sure remembers some of this
20:15:17 <decause> we tried very carefully to describe "it isn't just engineers, or just employees, but working with all kinds of people to advance open source, to benefit our customers."
20:15:26 <decause> we want better tech for the world, and to survive as an organization
20:15:42 <decause> If you work here, you know, but if you work for a diff kind of company, we think RHT is unusual
20:16:00 <decause> we include, in our code of conduct and ethics, we explicitly include a statement that says "yeah, we want you to work in Open Source"
20:16:24 <decause> we'd really like it if you kept RHT interests in mind, but if you see a conflict in RHT and project interests, you can do what is best for projects
20:16:28 <decause> that is interesting, and unusual
20:16:36 <decause> we expect our people to be ethical
20:16:44 <decause> doing what is right for communities and projects
20:17:02 <decause> me, providing guidance of where RHT needs to go, but sometimes, guidance isn't enough, or it isn't appropriate
20:17:13 <decause> the reason I talk about this, is it helps you understand RHT, the comapny
20:17:21 <decause> we don't expose this clearly to Fedora Audiences
20:17:28 <decause> #topic Fedora
20:17:38 <decause> this sometimes turns into flamewars on fedora-devel list
20:17:50 <decause> there is a certain amount of suspicion about "what is RHT doing?"
20:18:03 <decause> "never attribute to malice what could be explained by incompetance"
20:18:07 <decause> micommunication-r-us
20:18:16 <decause> internally even, we're not always aligned on what our goals are
20:18:39 <decause> my job, in management, is to help Eng. team udnrestand what goals are, and bring them to upstream comunities, where we contribute, all over the FOSS world
20:18:57 <decause> another thing you should know about RHT, unlike other folks who call themselves Open Source, we are *truly* about the upstream
20:19:04 <decause> it doesn't exist if it is not in the upstream somewhere
20:19:22 <decause> the reason some of these slides may look familiar, is because I cribbed them from the New Hire Orientation
20:19:38 <decause> you go through 3 days of that, because many who join us from closed orgs, they don't know how we work
20:19:49 <decause> I wanted to show the folks in Fedora, what we tell folks when they join
20:20:01 * decause can confirm, this is an actual slide from NHO as of March '15
20:20:14 <decause> hundreds of thousands of FOSS projects exist
20:20:20 <decause> RHT picks the ones we're going to participate in
20:20:43 <decause> we try to work hard in those communities, and eventually, all that ends up in things that we've hardened and tested, and will support and ship to customers, with a Red Hat label on it
20:20:49 <decause> that means *a lot* to Red Hat
20:20:56 <decause> it means we stand behind what we ship
20:21:21 <decause> but it also means there are many places where Red Hat has people participating genuinely, in those communities
20:21:27 <decause> if it isn't genuine, it won't fly
20:21:32 <decause> we want to be open, as a company
20:21:40 <decause> RHT + Fedora, Fedora + RHT, is truly open
20:21:45 <decause> we're not trying to hide things
20:21:56 <decause> we're disorganized sometimes, or there are differences in opinion
20:22:16 <decause> you've never seen 3-6 RHT people duking it out on the fedora-devel lists, right?...
20:22:19 <decause> *laughs*
20:22:32 <decause> maybe we shouldn't air dirty laundry so much in public, but I think it shows that we're tyring to be as open as we can
20:22:36 <decause> there is no big secret agenda
20:22:52 <decause> there are no midnight meetings where we say "how can we drop THAT into Fedora?"
20:23:15 <decause> "that won't work for Fedora, is there a way we can make that work in Fedora?" is what we say often
20:23:21 <decause> and it matters to RHT
20:23:32 <decause> we all succeed, and RHT succeeds, when Fedora Succeeds
20:23:38 <decause> we genuinely want Fedora to be popular
20:23:43 <decause> we want it to be useful to devs
20:23:46 <decause> available to customers
20:23:51 <decause> we want it to be wildly successful
20:23:54 <decause> the platform of choice
20:24:05 <decause> we want it to be the thing hardware partners ship as the default
20:24:22 <decause> #topic Goals for RHEL.next
20:24:33 <decause> this is where we talk about where RHT wants to take the direction of the OS
20:24:37 <decause> and things RHT would like to see in Fedora
20:24:49 <decause> some of these, you see RHT initiatives, that are worked on by RHT people
20:25:00 <decause> when I think about the next generation of RHEL, the OS, the Enterprise Platform
20:25:09 <decause> I keep hearing, and we keep seeing
20:25:14 <decause> "It is going to be modular"
20:25:30 <decause> it is going to be a small hardened core, with pieces layered, to control what interfaces look like
20:25:49 <decause> if you went to server talk, you see steve show how layers can proceed at their own speed, and still reliably work together
20:26:03 <decause> if you think about testing/validation/CI/CD these are all ideas that go together along with modularity
20:26:07 <decause> think back to Rings proposal
20:26:20 <decause> this is a theme RHT has been supporting and working twoards for some time now
20:26:31 <decause> this is Fedora, so there is no "OMG, we gotta do this NOW."
20:26:52 <decause> when I think about RHEL, and manageing the biuld deps and runtime deps
20:26:55 <decause> the amount of stuff to ship
20:27:01 <decause> I think we need to cut down on that
20:27:17 <decause> every package we ship, is a package that RHT the company has to support. If it is a build dep, do I really wanna ship it?
20:27:32 <decause> it does open the door to using it in ways taht weren't tested, so you can see the motivation
20:27:44 <decause> f23 has a focus on more security hardening
20:27:52 <decause> it does get in the way of ease of use
20:28:05 <decause> but hardening is becoming increasingly important for our customers, but also to Fedora
20:28:09 <decause> no one wants to get hacked
20:28:20 <decause> being easy to update release-to-release
20:28:27 <decause> what RHT sells, is not a version of RHEL
20:28:39 <decause> you don't buy RHEL6, then we you want to upgrade, you buy RHEL7
20:28:47 <decause> you subscribe, and then use whatever version you want to use
20:28:58 <decause> RHEL customers run a variety of versions
20:29:09 <decause> it means that when customers want to upgrade, they are not doing that in a vacuum
20:29:26 <decause> they aren't going to go in and throw out 10K RHEL6 servers, and then install RHEL7 10K
20:29:33 <decause> they want to upgrade what they have
20:29:42 <decause> we think about what that means for a sysadmin
20:29:55 <decause> making sure the commands they use are not going to destroy them on a new major version
20:29:58 <decause> think dnf/yum
20:30:04 <decause> lots of opportunity for lots of pain
20:30:10 <decause> for the folks who use yum
20:30:21 <decause> since many Fedora folks are sysadmins, we think this goal is reasonable in Fedora too
20:30:31 <decause> no one wants to learn a whole new commandline, but there is a name for that
20:30:45 <decause> the thing to think about, RHEL major release is usally like 6-7 Fedora releases
20:30:50 <decause> it is over a much longer timeframe
20:31:05 <decause> it is over a longer period of time, so we want compatibliity for longer peroids of time
20:31:15 <decause> Goal: automated compostion + automated testing
20:31:19 <decause> this is a generic goal overall
20:31:31 <decause> CI, continous integration, leading to continous dpeloyment
20:31:37 <decause> what htat means for an OS is anyone's guess atm
20:31:51 <decause> but we need better testing, earlier and earlier in the product creation cycle
20:31:57 <decause> we need to test all the way through
20:32:02 <decause> and from small to bigger and bigger units
20:32:22 <decause> so that when we do overall compose, the pieces are good, and we only fight integrations between elements bigger than packages
20:32:33 <decause> when you think about he work rel-eng is looking at, guess what?
20:32:41 <decause> there is a common theme. Every OS has the same problem.
20:32:54 <decause> this is work that we are trying to support in Fedora, and it feeds into what we do in RHEL
20:33:10 <decause> to think a bit about specific tech areas, (we are a tech crowd, i can't completely hand-wave)
20:33:15 <decause> RHEL is still evolving
20:33:22 <decause> Product management is talking to customers
20:33:31 <decause> they come back to eng teams, and say "what is emerging?"
20:33:40 <decause> which becomes "what upstreams do we need to make sure we ar esupporting?"
20:33:55 <decause> we have finite resources, so where do we encourage our people to work upstream?"
20:34:14 <decause> if you see us encouraging ipv6 support for utilities, tha tis in support of a RHEL.next theme likely
20:34:33 <decause> we're hearing from our customers, they want small core, they want rollback, they want system management that comes with atomic
20:34:40 <decause> we, RHT, are investing to make sure atomic has good support in Fedora
20:34:43 <decause> Python3
20:34:53 <decause> RHT and Fedora folks have been supporting this
20:35:04 <decause> we want python3 to be next major version for next major release
20:35:10 <decause> there are lots of other htings
20:35:17 <decause> we'd like to see more rolekit support for interesting roles
20:35:27 <decause> some things steve mentioned in his talk, alot o fthat, is work we're trying to support
20:35:40 <decause> I want to talk about the level of support that RHT provides for Fedora
20:35:43 <decause> I went looking
20:36:05 <decause> there are 35 full-time Red hat people--fedora infra, qa, rel-eng--35 all together, working full-time to support Fedora
20:36:29 <decause> if you assume we pay $50K/year, which, I'll tell you is a low estimate, that is 1.75M in salary every year, just to work on Fedora
20:36:37 <decause> that is not an inconsiderable investment
20:36:45 <decause> we have 100's of engs who package part-time
20:36:48 <decause> the COPR team
20:36:55 <decause> Fedora/EPEL/CoPR
20:37:05 <decause> amabassador money, hardware infrastructure
20:37:11 <decause> the hardware budget is my budget
20:37:21 <decause> Fedora got $144K of new hardware into your infra
20:37:28 <decause> I know we're trying to do the right thing here
20:37:32 <decause> we don't do this as a charity
20:37:48 <decause> if I told you we did this out of the goodness of our hearts, you wouldn't believe, and I wouldn't wanna lie
20:38:03 <decause> we do it because it helps biz, but it furthers an open source company with an open source miession
20:38:09 <decause> we believe stronly in FOSS
20:38:11 <decause> that is why we do it
20:38:15 <decause> it drives the economy
20:38:25 <decause> Fedora is more than just a beta cycle for what RHEL will become
20:38:33 <decause> it is a fountain of ideas where we make the world a better place
20:38:37 <decause> we wanna promote FOSS development
20:38:42 <decause> oh, and you know, sell RHEL
20:38:48 <decause> fortunately, they go together
20:38:51 <decause> we try to live this
20:39:03 <decause> we want to be part of a community that sees status quo as a hurdle, not a goal
20:39:08 <decause> being willing to change to improve things
20:39:17 <decause> not gratuitious, just for the heck of it change
20:39:20 <decause> but being free and open to new ideas
20:39:37 <decause> the distro where we just throw as many packages as we can, it's dead Jim
20:39:50 <decause> we want to support a constantly adapting/evolving community
20:39:55 <decause> or we go the way of the dinosaur
20:40:06 <decause> we want positive constructive dialog
20:40:08 <decause> we are not always right
20:40:13 <decause> we want feedback
20:40:16 <decause> we want people to share ideas
20:40:22 <decause> we want community participation
20:40:25 <decause> we want the community to grow
20:40:30 <decause> bigger communities benefit everyone
20:40:44 <decause> we want to be better about the way we communicate RHT goals/desires
20:40:53 <decause> if I look around, many faces here are RHT
20:41:01 <decause> you probably want to know "what does RHT want?"
20:41:07 <decause> internally, we're not always clear
20:41:11 <decause> we want to be more coordinated
20:41:32 <decause> we want to be clear about technical outcomes
20:41:39 <decause> but there must be argument, and hopefully the best idea wins
20:41:45 <decause> we can't dictate that, because we don't know
20:41:54 <decause> we gotta be understanding when things don't go our way
20:41:55 <decause> that is life
20:42:04 <decause> we will put our money where our mouth is, and send patches
20:42:11 <decause> I tried to show places where we do that already
20:42:22 <decause> some initiatives that red hat is working on, we gotta support what is happneing
20:42:27 <decause> #topic What does RHT Want?
20:42:39 <decause> To work better with Fedora to create the operating system for the next 10 years
20:42:52 <decause> we want to be Fedora, RHEL, and CentOS, because it is family.
20:43:00 <decause> #topic Q&A
20:43:15 <decause> Q: i'm glad you want to have better communication, and one voice. HOw does that look?
20:43:23 <decause> A: That is something we're working out
20:43:35 <decause> there are many senior tech folks here from RHT, who are also serious Fedora Contributors
20:43:52 <decause> when we discuss waht is next for RHEL, we will share with Fedora as Feature REquests
20:44:25 <decause> I try to do RHT talks about where are we going internally, and I will do those more often, to have RHT know the Fedora goals, to be more
20:45:06 <decause> Q: "To be clear about our desired technical outcomes" as a designer, this is almost too far from my perspective as a non-tech
20:45:14 <decause> Q: Understanding the use-case is a good way to do that!
20:45:28 <decause> A: Yes, if usecase isn't understood, or bigger picture, yeah, it won't work.
20:45:36 <decause> Doing that kind of communication is something that we need to get beter at that.
20:46:39 <mizmo> Q: (Remy) I've been a Fedora user/contrib for many years. went to new hire orientation in march, was optimistically skeptical. "we'll see how important the upstream is!" i can confirm, legit, those slides were in my new hire orientation in march
20:46:50 <mizmo> A: (Denise) Hey I'm management, I plagiarize
20:46:58 * mizmo hands mic to remy
20:47:04 * decause picks up
20:47:27 <decause> Q: sometimes the "secret" stuff, the tech outcomes, are secret, because you think you've communicated it, but you haven't
20:47:49 <decause> Q: There are some managers that don't do the best for the community. Can more be done to show communicating better?
20:48:10 <decause> A: sure, some product teams, are "heads-down" with due dates. Sometimes they forget what it is like to be upstream
20:48:15 <decause> sometimes it happens after the fact
20:48:28 <decause> and sometimes we need cooperation upstream, and we get into quandries
20:48:53 <decause> Q: No secret, people who come to work at RHT, don't come to work behind the scenes in close software.
20:49:30 <decause> Q: I've been a longterm contributor since Fedora1, and before Extras
20:49:44 <decause> I've been around for alot of the conspiracy theory flamewars, and in general, I dind' believe them
20:50:15 <decause> I can understand where a few come from, sometimes, it seems like ahuge effort is being spun up, but it is by an insular group of people, who know their goals, but no one knows the goals. it is confusing sometimes.
20:50:39 <decause> A: Yeah, that goes back to Langdon, when you forget that you're talking mostly to RHT'ers, and you forget that you're not including everyone all that often
20:50:49 <decause> I'm hoping that many discussions will happen at FLOCK/DevConf
20:51:17 <decause> Q: I'm going to assume the Fedora Council knows what the current goals are wrt the goals of RHT, and where the effort is going to go into.
20:51:23 <decause> A: Why would you assume that??? :P
20:52:19 <decause> Q: Docker, packageing guidelines, language guidelines... that effects a ton of stuff, and I didn't know. NOw ther eis a huge push, and it seemed to come out of the whole cloth. I didn't think it was such a big deal that the group would spontaneously organize. So long as the council knows, they can help us know
20:52:26 <decause> A: Yeah, regular council updates.
20:52:55 <decause> Q: RHT doesn' thave to follow Fedora, Fedora doesn't have to follow RHT, but is best when interests coincide. We can know where we're going, we can work together best.
20:53:12 <decause> A: Who gets to be the RHT spokesperson?
20:53:26 <decause> Q: mattdm, I report to Denise, and I talk to her very regularly.
20:53:41 <decause> Q: You can come to council meetings, and see what we think we know
20:53:45 <decause> (laughs)
20:54:10 <decause> Q: I remember when RHT was about 30ish people. We had devel list, where you found everything that was going on
20:54:22 <decause> as it was grown, scaling comms has been one of the hardest things
20:54:34 <decause> that isn't specific to RHT/Fedora, but to scaling all orgs of people
20:54:43 <decause> there is so much info that you can't pick out the important things
20:54:49 <decause> these are hard problems, there is nothing malicious
20:54:59 <decause> A: Point out where we screw up, yes :P
20:56:04 <decause> Q: Many RHT are Fedora Packagers. When they leave, we send an email to devel list, and say "we don't know what they want to do with them." someone can take or don't take them. it gives it back to the community.
20:57:15 <decause> Q: jzb, denise mentioned early RHT specifically said, I came from Citrix, which works on cloudstack, which competes with OpenStack. That is ok here, but not the case elsewhere. You go to Amazon, you can't necessarily keep working on Fedora Packages. If you go somewhere else, you may not have lattitude.
20:57:57 <decause> jzb++
20:57:57 <zodbot> decause: Karma for jzb changed to 3:  https://badges.fedoraproject.org/tags/cookie/any
20:58:47 <decause> Q: some people take it a step further, and make sure packages don't get abandoned. I hear that.
20:59:09 <decause> Q: jzb, yeah, we've reached info overload. I would love to see RHT invest in some resources to summarize discussions
21:00:15 <decause> #endmeeting