16:58:42 <klebedeff> #startmeeting
16:58:42 <chwido> Woof! Let's start the meeting. It's Wed Mar 16 16:58:42 2016 UTC. The chair is klebedeff. Information about me at https://wiki.openmandriva.org/en/Chwido.
16:58:42 <chwido> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:58:42 <chwido> You should add extra chair(s) just in case, with #chair nick.
16:58:42 <chwido> Have a good meeting, don't bark too much!
16:58:50 <klebedeff> #chair bero
16:58:50 <chwido> Current chairs: bero klebedeff
16:59:18 <bero> #topic Release status
16:59:39 <bero> So since I've been at a conference all of last week, couldn't keep track of everything...
17:00:05 <bero> but on my boxes Cooker seems to be working fairly well
17:00:14 <bero> fedya: do you have any updates on current release status?
17:00:28 <bero> fedya: in particular, did you get around to packaging the vanilla kernel yet?
17:01:47 <bero> ben79: crisb: any comments on release status?
17:02:14 <ben79> This since Saturday, I've been getting random desktop freeze ups, which seem to last exactly 60 seconds.
17:02:28 <bero> that's odd...
17:02:39 <Son_Goku> I’ve gotten freezes that last until I reboot :/
17:02:45 <Son_Goku> and it’s cross distro and DE
17:03:00 <bero> ben79: any idea which package (if any) you updated before this started happening?
17:03:23 <bero> Son_Goku: possibly a hardware issue?
17:03:31 <Son_Goku> the only commonality is kernel 4.4
17:03:34 <ben79> This is on hardware on a previously exixting Cooker and also on a fresh install from .iso # 100.
17:03:35 <Son_Goku> and of course, my hardware
17:04:05 <ben79> Don't have list of packages updated but I can get it and send an e-mail.
17:04:17 <bero> yes, that would be useful
17:04:32 <Son_Goku> haven’t tried Cooker yet, do we have kernel 4.4 yet?
17:04:50 <bero> ben79: anything interesting in dmesg or journalctl after it recovers from a freeze?
17:05:15 <bero> Son_Goku: no, some of the patches for the nrjQL patchset haven't been ported to anything later than 4.1.x yet, so cooker is on 4.1.18 right now
17:05:40 <fedya> bero kernel is semi-ready
17:05:41 <bero> We probably want to package a less patched 4.5 as well - needed for some hardware
17:05:42 <ben79> I don't see anything that explains freeze ups in journalctl or xsession-errors, didn't look at dmesg.
17:05:51 <fedya> i still droppping all ancient mandrake/mandriva crap
17:05:59 <fedya> is hard task
17:06:02 <bero> ben79: please look at dmesg next time it happens
17:06:25 <bero> fedya: I wonder if it wouldn't be best to just start over with a fresh linux-4.5 tarball
17:06:39 <ben79> bero: OK, will check dmesg and get list of packages updates after last Friday.
17:06:41 <crisb> is nicco not doing the kernel anymore then?
17:06:55 <fedya> bero i built vanilla kernel without special patches etc
17:07:02 <fedya> just as is
17:07:29 <bero> crisb: he is, we'll definitely keep the nrjQL kernel around. We just also need a more current, closer to upstream, version for hardware that isn't supported by 4.1.x well
17:07:47 <bero> e.g. on my laptop 4.5 works perfectly and 4.1 results in black screen, no audio, no ethernet
17:08:13 <bero> we need to support people with current hardware as well as people who want all the nrjQL improvements
17:08:20 <crisb> i mean cant he help with getting a 4.5 out
17:09:54 <bero> fedya: did you talk to him? My understanding (from 2nd hand bits and pieces) is that he's only interested in the nrjQL variant, but I may be wrong
17:14:41 <fedya> bero: i'm not
17:14:56 <fedya> i requested massive cleanup 2 months ago
17:15:04 <fedya> but he's bit depressed i think
17:17:33 <bero> yes, I've had that impression too
17:17:46 <bero> I wonder if there's anything we can do
17:17:56 <itchka> Hi
17:18:01 <itchka> Just back
17:18:03 <blackcrack> Hi itchka
17:18:52 <christann> regarding Ben79's problem, it could be due to an update to x11-server and related.
17:18:55 <itchka> About Nicco he's pretty upset that ROSA put anothe maintainer on the kernel.
17:19:26 <christann> I started to see similar strange things happen when running plasma.
17:19:35 <christann> only fails on real hardware, not Vbox.
17:19:58 <crisb> i see a lot of: https://sourceware.org/glibc/wiki/Release/2.23
17:20:01 <crisb> oops
17:20:09 <crisb> i mean
17:20:11 <bero> I'm running Cooker on real hardware (4 different boxes). One of them is headless, so I wouldn't see any UI freezes... The other 3 are fine
17:20:12 <crisb> libuwind: _unw_add_dynamic_fde: bad fde: FDE is really a CIE
17:20:19 <ben79> christann: possibly, as I recall there was an update of x11 packages right before problem revealed itself. Intel 4600HD graphics.
17:20:51 <christann> I have intel graphics as well.
17:20:54 <bero> Me too
17:21:05 <bero> Skylake on one, Broadwell on the other (third is radeon)
17:21:30 <ben79> My CPU is a Haswell
17:22:38 <fedya> bero see pm
17:23:45 <itchka> I have  machine her with intel graphics and Nvidia which continuously reboot. I was able to fix it in 2014.2 but it seems impossible to fix in cooker. Haven't tried the very latest iso yet.
17:27:58 <bero> how did you fix it in 2014.2?
17:29:03 <itchka> I rebuilt the initrd without the i915. I tried all the blacklisting tricks and none of them worked.
17:30:00 <itchka> It looks to me that if the i915 driver is loaded alongside any other drm driver some sort of system failure occurs.
17:30:48 <itchka> This is quite an old machine (64bit P4)
17:32:41 <itchka> I will do some more tests with the latest iso because even now I get it to come up with the nvidia driver only glxinfo among other this segfaults and I get a black screen. I've filed a bug.
17:34:17 <itchka> The only way the machine will come up cleanly is to use nomodeset.
17:35:00 <crisb> 915 doesnt see much love these days in mesa i think
17:35:39 <itchka> ceratinly seems that way.
17:35:54 <bero> I don't think that's a real i915 -- the module name is i915 even for Skylakes
17:36:04 <bero> and that gets all the attention in the world
17:36:25 <crisb> no thats 965
17:37:02 <bero> crisb: in mesa yes, in the kernel module no
17:37:34 <crisb> sure
17:40:49 <itchka> I think there is quite a big issue here. We have had people in the past who have reported blackscreens with dual graphics laptops which have Intel as the low power option and nVidia for games etc.
17:42:29 <itchka> Is our distro excluding people with modern hardware? Perhaps we should give this some priority?
17:45:18 <bero> itchka: that's exactly the issue I've seen on my new laptop before installing kernel 4.5 -- which is what prompted the kernel talk earlier
17:45:58 * n3npq suggests 1 paragraph for # s h a r e
17:46:12 <bero> at least in my case the problem was that the i915 kernel module from 4.1.x didn't recognize the Skylake GPU, so it loaded the nouveau driver instead, but the nvidia chip was disabled --> black screen
17:46:28 <bero> might be similar for you
17:47:21 <itchka> bero: It's not just nvidia either we hade a guy with 2014 who had a laptop with a two ATI card laptop that behaved in a similar fashion.
17:47:53 <bero> yes, that would probably be the same issue -- external chip disabled, Intel chip not recognized by old driver -> black screen
17:49:45 <christann> I have one graphics card - and Intel one. The problem that I am seeing is when I run Plasma, kded5 and kglobalaccel5 crash
17:50:19 <bero> not seeing that on any of mine
17:50:52 <christann> LXQT is okay
17:51:34 <itchka> christann: That's pretty much what I'm seeing. If you go to a terminal you'll see that glxinfo crashed plus a few othwer things like kdeconnectd
17:52:12 <itchka> you can see them if you use journalctl from a dummy console.
17:52:31 <crisb> itchka christann: but what intel gpu is it
17:52:55 <itchka> criab: Dunno let me check
17:53:54 <christann> GPU XeonE3-1200 CPU i7-4770
17:54:59 <christann> I am running kernel 4.1.15 server (per fedya).
17:55:43 <christann> I have to go.
17:56:33 <crisb> itchka: just cat /proc/cpuinfo | grep "model name"  will be enough
17:56:39 <itchka> crisb: Mine is Dell 821915G/GV/910GL
17:57:56 <crisb> ah so that is the old 915 driver for mesa
17:59:45 <itchka> This is quite an old test box.
18:00:54 <bero> mesa 11.2 should be released any day now, so we should probably update the 11.2 snapshot we currently have - but I doubt it'll have any fixes relevant for this
18:01:09 <bero> if this was caused by Mesa, chances are you'd still see a UI (just no acceleration)
18:01:19 <bero> this is much more likely an issue with the kernel driver
18:01:36 <bero> so I'd think the 4.5 kernel is more likely to fix it
18:02:40 <itchka> My own conclusion which I cannot prove indicated that interaction between glxinfo and nouveau was causing a segfault.
18:04:11 <bero> ok, will update the mesa snapshot today so we can see if that fixes anything
18:04:23 <itchka> If there's no glxinfo fro kf5 to work with all the other failure were the likely result.
18:04:26 <bero> but we'll need to update it again once the 11.2 final is officially released
18:06:26 <bero> #share Several people have seen black screens on bootup (hardware dependent issue), will try getting Mesa closer to 11.2 final and adding a 4.5 based kernel to see if this is already fixed upstream.
18:06:35 <bero> Do we have anything else on release status?
18:07:12 <bero> we do have sub-points on Qt 5.6, Plasma 5.6 and Freezing and branching, but I'd like to wait with those until TPG gets back
18:07:33 <itchka> There is one other strange thing happening for which I cannot account. When my screen locks on this box sometimes I get the sdddm locker and others I get the xscreensaver locker. I'm pretty sure that this is not supposed to happe :)
18:07:42 <bero> I think we have some things on AoB today, so unless there's objections, let's go to AoB first
18:08:23 <bero> itchka: that's indeed a bit weird. Probably somehow you have 2 screensavers running and whichever gets the signal first wins
18:09:32 <itchka> bero: Just flagging it for future examination. This was a standard install.
18:09:43 <itchka> Let's do AOB
18:09:49 <bero> #topic AoB
18:09:56 <bero> tapwag: I think you wanted to talk about CeBIT
18:11:00 <tapwag> bero, Yes
18:12:25 <tapwag> Okay, I have been to CeBIT in Hannover today and got a full-event ticket from IBM. As a matter of courtesey IBM was the first stand I visited and I hat a look at their POWER systems which are quite impressive and their sponsoring projects I will now speak about
18:13:55 <tapwag> So what is the deal? IBM offers two programs called BlueMix and Softlayer, the latter being a Metal as a Service solution and we could be elegible for 1000 US-Dollars/year worth of processing time in their IBM cloud
18:14:33 <tapwag> The good thing is that we can turn the cloud on and off. So if we need to build an ISO we can turn it on, build it and then turn it off again
18:15:23 <tapwag> The person I spoke to gave me the following link: developer.ibm.com/startups - they said that non-profits are also elegible
18:15:50 <blackcrack> [suggestion] maybe a automatich turnon/turnoff script...[/suggestion]
18:16:47 <tapwag> I got two business cards. One from a person dealing with POWER and OpenPOWER Foundation who is currently writing his thesis with IBM and the other one from a person who is the lead for this programme
18:17:13 <tapwag> I am not going to send an application. This would be Council's job
18:17:43 <tapwag> And again this is not meant to sabotage the efforts that we made with the new ABF
18:17:54 <tapwag> That's pretty much it
18:18:00 <fedya> https://github.com/docker/docker/issues/20950
18:18:08 <fedya> another one issue with docker and kernel
18:18:32 <bero> as far as I'm concerned it actually goes well with the new ABF -- the new ABF makes it easy to add new nodes hosted elsewhere
18:18:47 <tapwag> I was fairly impressed with the POWER server with 1 TB of RAM
18:18:56 <fedya> well latest stable version is 4.1.18
18:19:00 <tapwag> costs 10000 Dollars
18:19:01 <itchka> tapwag: Looks like there's a 1 month free trial.
18:19:38 <bero> That's not actually so bad compared to an ultra high end x86 box
18:19:41 <tapwag> itchka, What they told me is that they would give us 1000 US-Dollars worth of credit and we would pay on how we use it (cores, minutes etc.)
18:20:14 <tapwag> 8 cores and each one of them can apperently do 4 threads
18:20:17 <tapwag> IIRC
18:20:27 <bero> but I doubt we can support another architecture without extra manpower
18:20:44 <tapwag> Agreed
18:21:13 <n3npq> perhaps a proposal might attract some man power?
18:21:30 <tapwag> I bought a PowerPC G3 on ebay today ;-)
18:21:47 <tapwag> But I am also completely with n3npq
18:22:01 <n3npq> tapwag: I'd also chat up Fabian Arrotin in Centos … he's a really really nice guy and knows lots about POWER and IBM
18:22:25 <fedya> reboot
18:22:27 <tapwag> But after all: At OpenMandriva we are focussing on the Desktop. At least this it how it has been
18:23:10 <tapwag> Just wanted to put the issue of going with IBM on the table
18:23:20 <n3npq> tapwag: yes the OMA focus is Desktop, but the infrastructure on POWER and the build systems are common elements
18:25:03 <blackcrack> [suggestion] if we use the offers, should we use it as big news.. not anyone become some offer..[/suggestion]
18:28:18 <tapwag> blackcrack, I have been thinking about that as well. But eventually other distributions might or will copy this
18:29:14 <blackcrack> well, we are the first one ;) tapwag :)
18:29:18 <tapwag> IBM invests a lot of money and they also had the IBM "LinuxONE Rockhopper" there as well
18:29:37 <tapwag> a lot of money in Linux that is
18:29:49 <itchka> I'm not clear here bero are you saying we would need to support the POWER architecture in order to use these servers for abf.
18:29:55 <tapwag> Even though they are laying off quite a lot of staff at the moment
18:30:07 <blackcrack> brb
18:30:24 <tapwag> As far as I understood it we do not need to support POWER
18:30:54 <tapwag> We can just use their processing power and they have a lot of it. Think of the Watson supercomputer
18:31:18 <n3npq> tapwag: if it helps, I can try doing rpm development on an OMA sponsored POWER project. I really don't care what CPU or modern POSIX I use. With a working rpm, getting Mandriva there (at least in a chroot) is just a boot strapping operation.
18:31:55 <ben79> need to reboot to get in to Cooker/OMLx3/2015 (could we once and for all decide to either call it '3' or 2015?)
18:31:58 <tapwag> I did the negotiations as fairly modest saying that we have already have a build farm but we could use a lot more
18:32:10 <bero> What do you want to run on it? The thing we need processing power for is primarily building packages, and I don't think we want to start crosscompiling x86 packages...
18:32:36 <bero> n3npq: that could be interesting... Certainly not opposed to taking patches to support it, we just need to make sure we don't overload current maintainers with an extra arch
18:32:36 <n3npq> note that I have no idea what "Desktop" means on a 1Tb POWER
18:32:43 <tapwag> bero, Building packages and hosting them as well
18:33:08 <crisb> but is the free cpu time on power nodes only
18:33:57 <tapwag> n3npq, I asked them if they would also be producing consumer mainboards such as the SAM for the AmigaOne which is also based on POWER but from what I have been told they have no interest of producing them.
18:34:41 <tapwag> Believe it or not: Amiga is still around www.amigakit.eu
18:34:51 <n3npq> bero: Poky/Yocto cross compiling might be useful if on POWER. there are benefits to participation with other projects
18:35:41 <tapwag> And after all there aren't that many distributions supporting POWER
18:36:03 <tapwag> One that comes to my mind is YellowDog Linux
18:36:06 <crisb> thats because its a very niche arch
18:36:10 <n3npq> tapwag: note that "Desktop" is largely meaningless on ARM IoT as it is on Big Iron POWER
18:36:27 <crisb> and barely anyone has hardware
18:37:15 <tapwag> n3npq, I agree. It was just that at OpenMandriva we decided to do a desktop distribution
18:38:05 <n3npq> yes and there is nothing wrong with a Desktop focus: OMA has the best KDE around.
18:38:06 <Son_Goku> isn’t there that Talos workstation using POWER8?
18:38:07 <tapwag> crisb, After all they told me that they aren't interested to produce consumer mainboards
18:38:33 <tapwag> The rack that I have been shown was a POWER8
18:38:57 <Son_Goku> tapwag: off the top of my head, Fedora, CentOS, openSUSE, and Debian still support POWER
18:39:05 <n3npq> Son_Goku: there's lots of POWER on game consoles, just jail breaking is often painful
18:39:15 <Son_Goku> yeah, but that’s going away too
18:39:28 <tapwag> Son_Goku, At least with openSUSE it is mainly 64-bit POWER
18:40:01 <Son_Goku> Fedora switched off building 32-bit POWER back in 2012 or so
18:40:19 <Son_Goku> Debian’s 32-bit POWER is basically broken
18:40:41 <tapwag> Anyway, the question that I would like to address to itchka and council if OpenMandriva is interested in having IBM as our MaaS/Cloud etc. provider.
18:40:51 <bero> yes, 64-bit is the more interesting variant anyway
18:40:58 <tapwag> Feel free to discuss this
18:41:23 <n3npq> tapwag: anyways, my offer to get "working" rpm on POWER8 is open if that helps setup a project. And Poky/Yocto can cross-compile onto POWER last I checked and also use rpm5
18:41:53 <Son_Goku> aside from the Talos workstation, I don’t see much in the way of POWER outside of servers
18:42:07 <blackcrack> .
18:42:15 <tapwag> n3npq, Sounds very good to me but I am not in council and I am just bringing the offer from IBM forward
18:42:22 <n3npq> Son_Goku: correct. and why most distros are dropping POWER
18:42:49 <Son_Goku> that said, I wish I could play with a POWER8 computer :/
18:42:58 <Son_Goku> I’ve heard good things about the architecture
18:43:28 <tapwag> Son_Goku, In ten years time if things keep going the way they went you might have this at your wristwatch at some point in time
18:44:09 <bero> so the interesting question there is mostly whether or not we have a use for CPU time etc. on a power box when we don't have a port
18:44:15 <tapwag> Anyway, the issue is not to discuss the pros and cons of POWER but if OpenMandriva is interested in partnering with IBM
18:45:17 <tapwag> bero, IBM never said to me that in the programme they are offering we would have to use POWER-systems
18:45:28 <tapwag> I just mentioned that I found it interesting
18:45:57 <n3npq> tapwag: do you know pcpa? you might talk to him as well, and he might be interested in POWER
18:46:01 <tapwag> They just said to me: Apply for it, you get 1000 US-Dollars worth of credit and you can blow that on the IBM cloud
18:46:43 <tapwag> That's what they suggested. The stand is always fairly big. In fact IBM still owns part of the halls at CeBIT
18:46:47 <bero> hmm, if that can be on something we support, I guess we could set up another build slave (and just make sure it shuts down after using up $1000
18:47:33 <tapwag> bero, What I have been thinking about: TPG decides to build an image, he turns on the cloud, builds the image and turns it off again
18:47:59 <n3npq> tapwag: if you (and OMA) wish, a collaboration between rpm5.org & openmandriva.org as a proposal might be possible. there's a couple of people @rpm5.org who might be interested in POWER8 access
18:48:02 <tapwag> 1000 US-Dollars/year they told me
18:48:22 <n3npq> and if it all crashes & burns you can blame rpm5.org ;-)
18:48:31 <Son_Goku> awesome
18:48:39 <Son_Goku> blame rpm5 for everything ;P
18:48:52 <n3npq> all ter software belongs to rpm
18:48:56 <n3npq> yer
18:48:57 <tapwag> n3npq, Unfortunately I can't afford such a machine but I am in fact very glad and proud to have you onboard with us
18:50:22 <itchka> tapwag: I can't see what IBM are going to get out of this from use except perhaps some work that might improve docker on their architecture.
18:51:07 <bero> publicity and mindshare
18:51:39 <bero> it's not like it actually costs them anything to provide their services
18:51:41 <itchka> I go the impression that when we looked before what they really wanted was for people to develop applications for the arch which is not atall what we would be doing.
18:51:48 <n3npq> does OMA have any concept of a SIG? getting a few people together to try-and-see within a SIG need not suck up scarce man power. pcpa used to borrow my G5 to work on his power jit compiler, but he likely has better resources through @redhat.com (but a different justification is needed with a boss than as a participant in a project)
18:53:12 <n3npq> itchka: yes IBM is seeding their own infrastructure and wishes $$$ applications/support.
18:53:30 <bero> n3npq: Not currently/officially, we're so few people that we can't really have any SIGs... But in a way probably fedya and I are an ARM SIG
18:54:19 <itchka> You could almost view our new abf as a SIG.
18:54:37 <n3npq> but -- say -- M$ has just announced SQL Server on linux. If M$ means source code, then getting SQL Server packaged up starts to have enoufgh branding buzz words that there is interesting PR
18:55:00 <n3npq> bero: include me in your ARM sig anytime
18:55:23 <bero> I think it's binary and for-pay, but didn't check closely
18:55:50 <bero> n3npq: There's no special SIG meetings or anything, so feel free to count yourself in ;) essentially all it means is we're doing most of the work on ARM porting
18:56:11 <HisShadow> what does SIG stand for
18:56:39 <blackcrack> exactly, this is my question too *bG*
18:56:43 <bero> HisShadow: Special Interest Group
18:56:53 <n3npq> bero: I'd *LOVE* to be able to use rpm natively. you have no idea what pain there is in --with-path-versioned parallel installs
18:57:29 <n3npq> sadly, I still cannot type "make install" on cooker without spending days trying to figger out WTF?!?
18:57:53 <n3npq> Fedora/Centos are eually poisonous
18:57:58 <n3npq> equally
18:58:27 <ben79> bero: On Cooker ML see : Re: [OM Cooker] Desktop freeze up return with a vengeance.
18:59:00 <itchka> I think we are drifting off the point a bit. I think we need to answer tapwags question.
18:59:13 <bero> n3npq: what's the problem? I haven't had problems with "make install" or the likes lately, but maybe I'm just so used to the specifics of the system that I no longer see them
18:59:18 <tapwag> itchka, You don't need to answer this right now
18:59:35 <tapwag> I would rather like to have this discussed in Council
18:59:54 <blackcrack> +1
19:00:12 <tapwag> What I am saying is that IBM is giving away resources that OpenMandriva might want to tap into
19:00:42 <tapwag> If you want to make use of their offer I personally think it is a good idea
19:00:48 <n3npq> bero: I mean "make install" from a CVS tree … meanwhile there are so many patches and configuration that nothing "works"
19:00:49 <bero> I think we need to find out more (e.g. if it has to be power boxes etc.)
19:00:51 <tapwag> Not all members might agree on this
19:01:08 <itchka> To be honest tapwag it's as much a technical issue as it is a council one. It would be pointless for council to approve and apply if the developers cannot take advantage of what is on offer.
19:01:13 <tapwag> So we might want to cast a community vote as well
19:02:08 <tapwag> itchka, They never said that it would have to be POWER. It was just that I mentioned that the platform looks very good to me.
19:02:14 <n3npq> I had to sit on an 2011 Mandriva system for 2 years with no ability to upgrade to preserve mdawkins chroot building … and its the chroot that screwed the ability to upgrade the host mandriva system
19:02:30 <tapwag> The stand was quite huge and there were lots of people to talk to
19:03:18 <n3npq> bero: but if you can do "make install" from rpm5.org CVS without breaking the entire URPMI stack, then I am much more likely to run/develop on OMA
19:03:36 <Son_Goku> n3npq: that’s probably not going to happen
19:03:40 <tapwag> Anyway, I strongly believe that IBM has more computing power than ROSA had or our own infrastructure has now
19:03:55 <Son_Goku> even on the rpm.org side, we still have to patch several behaviors in rpm to make urpmi not break
19:04:14 <Son_Goku> *on the Mageia side
19:04:20 <bero> n3npq: ok, haven't tried that in a while... but would be nice to make sure it can work. Probably means we'll have to get some not-so-nice patches into cvs
19:04:32 <n3npq> Son_Goku: what isn't going to happen? Typing "make install" on OMA preserbving the ability to upgrade?
19:04:47 <Son_Goku> yes
19:05:08 <Son_Goku> admittedly, you’ve already adapted rpm5 more than it ever should have for urpmi
19:05:47 <n3npq> you do realize that I am the rpm maintainer of record in OMA as well as the "upstream" maintainer? And you are telling me that those two roles are incompatible? be serious ...
19:06:32 <tapwag> itchka, So why not tap into what IBM has on offer? I can see that there might be a lock-in effect but other than that I can't see any drawbacks
19:06:58 <itchka> tapwag: I have no doubt about that from what you say one processor could support two rpm builders whereas to do the same on most of our current nodes require 8 processors. So just one of those servers would make an enormous difference to abf massbuild times. The problem is how much time has to be invested in order to gain that advantage.
19:07:13 <fedya> n3npq if i do for you one-button build for RPM5 are you become happy?
19:07:16 <n3npq> bero: or remove some not-so-nice patches from OMA … meanwhile I am in zugzwang with attacks from both roles.
19:07:44 <tapwag> itchka, n3npq once mentioned that it took him about 6 months to get this approved if I am not mistaken
19:07:45 <n3npq> fedya: we have different meanings for "RPM5"
19:08:24 <Son_Goku> n3npq: I’m saying that in order to preserve your sanity when it comes to maintaining rpm 5 as upstream, it’s going to be really hard to take in oma specific patches
19:08:24 <bero> n3npq: Probably comes back to "urpmi must be replaced" -- there's just too many dirty things in there that nobody really knows how to fix
19:08:36 <n3npq> I can (and have) built rpm5 on 14 Mandriva-derived OS's in buildbots and ran that for months.
19:08:40 <fedya> yes, need to replace urpmi ;)
19:08:42 <tapwag> After all IBM is a cooperation with lots of bureaucracy
19:09:00 <bero> Son_Goku: We want to get rid of most of those...
19:09:10 <itchka> tapwag: I was just about to ask and you have explained.
19:09:32 <bero> just might need to continue some kludges like distepoch
19:09:40 <Son_Goku> or kill it with fire
19:09:46 <Son_Goku> it’s actually possible to go without it
19:09:50 <n3npq> Son_Goku: I'm here to merge the differtences, not to do "rpm therapy" and be told how speshul the OMA patches are
19:10:08 <Son_Goku> n3npq: I just worry about your sanity, Jeff
19:10:23 <Son_Goku> I don’t want you to have a meltdown because of eye-scalding badness
19:10:55 <tapwag> itchka, Just one more thing: The offer from IBM is scalable you can add more cores, more memory etc. and turn it on and off again just like water
19:11:18 <n3npq> Son_Goku: so send sympathy cards. I have a rpm-5.4.16 release underway, and have tried for 2+ years to give OMA an upgrade path
19:11:36 <tapwag> itchka, So the idea is: We do an ISO Build, turn on the cloud, have it build and then turn it off again
19:12:05 <itchka> tapwag: I understood. This makes it very suitable for the current abf architecture.
19:12:41 <blackcrack> +1
19:13:00 <blackcrack> with 4 TB Raid, indeed yes ..
19:13:02 <tapwag> itchka, Other than the lock-in effect I can't see any drawbacks. And this would be a discussion for Council if OpenMandriva wants to engage with a partner who has more financial resouces than we have
19:13:03 <itchka> tapwag: Isobuilds are a small thing compared to mass builds. It's for mass builds that we need the real power.
19:13:07 <Son_Goku> n3npq: I would, but I don’t know where to send them ;)
19:13:38 <bero> I'd also rather use it for package builds than iso builds (iso builds need I/O, but almost no CPU power)
19:13:57 <bero> the question is how many builds we can fit into $1000
19:14:18 <itchka> bero: Feasability study?
19:14:35 <n3npq> Son_Goku: then I will send you my sympathies when I get to doing LCOV/GCOV coverage tests and coverity scans on the 500+ Kloc that I maintain that OMA cannot use
19:14:39 <bero> guess so -- we'll need to figure out just how much those $1000 would get us
19:14:44 <n3npq> this week
19:14:48 <tapwag> bero, There would be several factors involved: How many cores do we assign to this build, how much memory, how much time
19:15:38 <bero> Let's say we configure it like a decent build box -- 8 cores, 64 GB RAM, time - well, as long as it takes to finish the build (with shutdown as soon as we reach $999)
19:15:41 <tapwag> Anyway, it is a suggestion that you can make use of.
19:15:44 <bero> the question is how much time it would get us
19:16:06 <tapwag> bero, As far as I know you can turn your instances on and off
19:16:15 <itchka> the question is first how many builders.
19:16:28 <tapwag> At least this is how it works with Amazon Web Services AWS
19:16:33 <fedya> bero 300$ in google cloud with 6CPU and 8GB RAM is for 2 months
19:16:37 <itchka> fedya: how many threads needed per rpm-builder?
19:16:44 <bero> tapwag: yes -- pretty sure that's how it works
19:16:52 <fedya> itchka 8 cores per 2 builders
19:16:59 <fedya> 4 core per builder
19:17:13 <fedya> possible less, but only for ix86 builds
19:17:55 <fedya> googlebuilder.openmandriva.org is free trial build node in google cloud
19:18:02 <itchka> fedya: so evenif each core can run more threads we still nee four cores per builder?
19:18:02 <fedya> they gave 300$
19:18:18 <itchka> s/nee/need
19:18:34 <fedya> itchka: what is 'more threads' ? ;)
19:18:40 <tapwag> fedya, IBM was quite shocked when they heard that Google also has a supercomputer now
19:18:59 <tapwag> At least that was the impression of an HR person I spoke to
19:19:05 <fedya> proyvind promised big bertha to me
19:19:14 <tapwag> Innovation cycles are getting shorter
19:19:34 <fedya> but looks like he doing things in mageia now (or sabotages)
19:19:37 <n3npq> fedya: if googlebuilder.openmandriva.org is mostly functional, that is a candidate for # s h a r e … I would attempt to use even if noone else does
19:19:43 <tapwag> So IBM can't say: We built Watson so we can lay back for the next couple of years
19:20:20 <tapwag> Just one Off-Topic question: Who is competing in that Human-vs-Machine Go-competition right now?
19:20:20 <fedya> n3npq you want ssh into googlebuilder?
19:20:57 <itchka> fedya: I'm not exactly sure IBM told tapwag that each of their processors can run 4 threads which I presume means something like 4 different processes but I do not know for certain.
19:20:59 <tapwag> The last results I heard off was that the computer is 2:0 ahead of the human play
19:21:00 <n3npq> fedya: I (and perhaps others) want some reasonable description first
19:21:14 <fedya> about what?
19:21:44 <tapwag> itchka, That was when I visited the "Systems" area of their stand
19:22:46 <tapwag> Again, the stand was really huge and IBM still owns part of the Hannover CeBIT premises
19:27:43 <Son_Goku> fedya, Big Bertha is dead in RMA land
19:28:32 <fedya> 2671 strings vs 1253 strings! in new kernel spec!
19:28:40 <fedya> Son_Goku sad
19:29:00 <Son_Goku> it broke a few months after he set it up, and he tried to get it RMA'd
19:29:10 <Son_Goku> afaik, he still hasn’t gotten the replacement parts yet to put it back together
19:29:27 <tapwag> itchka, If we do accept their offer I am assuming klebedeff would have to write to IBM to apply for this sceme
19:30:27 <tapwag> But I do have the impression that they are fairly liberal with their resources. Maybe "Ginny" wants to bring the whole Linux community under her control, who knows?
19:30:54 <tapwag> That's why I want to have this discussed in Council
19:31:11 <itchka> tapwag: Certainly somebody will have to write to them.
19:31:53 <tapwag> I spelt her name wrong: https://de.wikipedia.org/wiki/Virginia_Rometty
19:32:47 <tapwag> Well, could be that she wants all distributions signing up to her plan. I never met her and so I don't really know about her plans
19:32:54 <itchka> tapwag: Lets find out a bit more and see where we go from there.
19:34:28 <n3npq> fedya: about googlebuilder.openmandriva.com … I am assuming that is an ABF node of some sort, but perhaps is an alternative means to build packages for cooker
19:34:58 <n3npq> without a description I don't know.
19:35:46 <bero> n3npq: It's an abf build node -- essentially when you throw a package at abf to be built, it'll end up being built either on one of our machines or in googlebuilder (whichever is available and not busy)
19:36:06 <klebedeff> tapwag please summarize the offer and send it to council ML
19:36:37 <fedya> n3npq right this server building packages for cooker and omv2014
19:36:48 <bero> and try to find some details (like just how much CPU time we'd get for $1000)
19:38:35 <n3npq> bero: the usage case would be trying to add my blades to ABF. the difference is f23/Centos7 hosting. seeing examples of how nodes are set up helps a lot. and the starting point would be a # s h a r e (or a deployment plan)
19:42:37 <bero> n3npq: It's just a matter of running a few docker commands these days
19:42:59 <bero> fedya or HisShadow: would you give the exact specifics to n3npq and # s h a r e them? You know best how to do it...
19:43:41 <fedya> bero i did it months ago
19:43:51 <fedya> i dunno what i can do more here
19:44:29 <fedya> i even made screencast https://asciinema.org/a/9c5mzq43h15kmeg4roiq8yvok?autoplay=1
19:44:40 <fedya> https://github.com/OpenMandrivaSoftware/docker-builder
19:45:10 <n3npq> fedya: sorry for not being informed: I am late to this party
19:45:11 <fedya> what need? any modern linux distro: ubuntu, fedora, centos7, coreos, etc and overlayfs.ko in kernel
19:45:30 <fedya> nothing more
19:46:01 <n3npq> so what remains to add hi.jbj.org as an ABF build node? that is/was the goal since November
19:47:29 <tapwag> I would like to point out that the discussions I had with IBM are not meant to sabotage anyone's efforts or put anyone under pressure. It's just meant to make you aware of an offer that is there.
19:48:37 <fedya> n3npq well i lost login credentials to one of your servers where runnning cockpit
19:49:24 <blackcrack> well Tabwag, send the buendle of informations to the consil and we whant see .. they should speak about in the Consil.
19:49:33 <blackcrack> tapwag:
19:50:16 <n3npq> try ssh -p 19022 jbj.org (if IPv4) or start thinking about IPv6 … I have only a single assigned IPv4 address which maps to OS X server
19:50:50 <fedya> ah got it
19:51:16 <fedya> rancher server running there
19:51:46 <n3npq> yes … kill it if you wish. poke me if you need a reboot however
19:51:57 <fedya> n3npq i can add this server as build node?
19:52:08 <n3npq> fedya: yes … please ;-)
19:52:12 <fedya> ok
19:52:40 <n3npq> I was running ranc=her server to not interfere with your ABF maintenance/development
19:53:14 <tapwag> blackcrack, klebedeff Okay, will send this to the council-ml tomorrow
19:53:38 <itchka> tapwag: Thank you
19:53:41 <blackcrack> include prozessortime CPU cousts and so on :)
19:53:45 <n3npq> if I can see a working instance, I have several blades and several VM's that can be used
19:54:05 <blackcrack> what ask's bero too ..
19:54:50 <fedya> n3npq wait a bit, i will add it right now
19:55:50 <tapwag> I will do a summary. The person that I am worrying about is fedya as he spent a lot of time on our independent infrastructure. fedya It's not personal or anything.
19:56:03 <Son_Goku> arrgh
19:56:04 <Son_Goku> we’re almost out of space on jbj.org
19:56:27 <tapwag> At least this was the impression I got from recent IRC conversations
19:58:41 <tapwag> Also itchka liked this idea and I am a strong supporter of having this independent infrastructure up and running the way it does at the moment
19:58:42 <blackcrack> tapwag: ask, how we should make this with the harddisk for our ABF .. fedya .. how.. itchka tell..
19:58:48 <Son_Goku> n3npq: somehow, 96% of the root filesystem is used up
19:59:17 <blackcrack> 2x 4 TB
20:00:21 <itchka> blackcrack: I don't understand you have 2 x 4Tb drives free?
20:00:50 <blackcrack> i whant pay with Maik and  an Anonymus 300Euros for our ABF
20:01:07 <blackcrack> 2X 4TB' Red, WD
20:01:35 <n3npq> Son_Goku: lemme look …
20:01:36 <blackcrack> this schoul be included directly into the abf
20:01:52 <blackcrack> as raid
20:02:04 <Son_Goku> n3npq: copr crashed due to out of space
20:02:08 <Son_Goku> among other things
20:02:29 <blackcrack> becaue.. out of the space.. indeed *bg*
20:02:34 <blackcrack> because..
20:02:59 <itchka> blackcrack: So you have a server with 2 x 4Gig raid?
20:03:20 <tapwag> itchka, He is willing to buy two 4 GB hard drives for us
20:03:44 <tapwag> He just wants to make sure that they are implemented properly so everyone benefits
20:04:12 <blackcrack> itchka:  nope, i whant move 2x 4TB's into the OMLx ABF
20:04:20 <HisShadow> 4 GB, nice, we could move to Windows 2000 finally :P
20:04:31 <fedya> :D
20:04:50 <blackcrack> Terabyte ..
20:04:59 <tapwag> Oh sorry, it was 2 4 Terabytes of storage
20:05:01 <blackcrack> *lol*
20:05:19 <blackcrack> and it is for Win 3.11
20:05:24 <blackcrack> *roflmao*
20:05:41 <blackcrack> no, in serous, 2x 4TB for the OMLx ABF
20:06:09 <itchka> blackcrack: Ok I'm not sure how we can do this. Rather than buy drives it might be more effective to add extra space to our existing two servers.
20:06:17 * tapwag appreciates blackcracks willingness to buy storage for OpenMandriva
20:07:14 <blackcrack> where can hande that to move the 2 harddisk's into the Server .. like i have become with, Bero, this computer stay in the swiss ?
20:07:59 <blackcrack> the OmlX ABF ?
20:08:41 <itchka> blackcrack: The servers that were in Switzerland are no longer.
20:08:48 <tapwag> If we have someone willing to buy this for OpenMandriva, please have it integrated
20:09:05 <itchka> we are using cheap servers from soyoustart
20:09:46 <blackcrack> :\ and how can we do this for make 2X 4TB into it.. it is possible ?
20:09:47 <itchka> tapwag: It a practical problem that we cannot easily solve.
20:10:00 <fedya> i remeber that we have own servers
20:10:05 <fedya> where it is?
20:10:56 <tapwag> From what I have read in the pro-linux.de news we made ourselves indepented and I thought the main idea was getting decentralized...
20:11:08 <itchka> fedya: They are in Paris  at the moment sitting in a corner in JCL's wifes office.
20:11:47 <blackcrack> the abf ?
20:11:58 <fedya> itchka: but why/
20:12:00 <fedya> ?
20:12:03 <itchka> We are not using them because it's more expensive to host your own servers than to rent them.
20:12:47 <fedya> itchka: what the f* i'm reading
20:13:01 <fedya> itchka: collocation is more expensive than rent?
20:13:02 <fedya> oh
20:13:51 <fedya> here is collocation per months cost 30 EUR with 100mbit channel and unlimited traffic
20:14:14 <itchka> soyoustart servers cost us around 25 Euros a month we could not host our own for that in Europe.
20:15:04 <klebedeff> itchka let's discuss blackcrack 's initiative in AoB
20:15:22 <fedya> collocation in Selectel 35Eur/month
20:15:46 <itchka> That's why I say it would be better to use the money that Blackcrack is offering to beef up the soyoustart servers with addons
20:16:06 <blackcrack> we are in AoB.. isnt it ? :\ take a look on the Chan destription ..
20:16:11 <fedya> average price ~32 eur
20:17:54 <blackcrack> it was out of space where was build packages, this was the reason why i have told, letz buy a 4TB disk.. or 2 as rayd .. and i have ask there in the chan, if there be space for setting up our servder
20:18:27 <itchka> fedya: Then there is maintenance etc, etc I'll bet that's a base price. In my view anyone passing through Paris should pick one up and take it home and add it to abf as a node.
20:18:50 <fedya> itchka: indeed
20:18:52 <blackcrack> ehh server
20:19:55 <itchka> They are certainly more modern than my old  HP ML350 G4
20:20:42 <itchka> blackcrack: Have you bought these drives yet?
20:20:54 <fedya> ashledombos ^^
20:21:54 <blackcrack> no, i'ts on the start, the money is now on my bank and i wait, how we make it..
20:23:00 <itchka> blackcrack: Thank you for your kind offer we need to look at the best way of taking advantage of it.
20:23:12 <blackcrack> oky..
20:23:30 <klebedeff> itchka shall we discuss now?
20:23:30 <blackcrack> stay tuned
20:24:48 <itchka> klebedeff: I'll talk to Blacky about it and we will try and find a way to do it.
20:25:09 <klebedeff> so we take it forward next week?
20:25:43 <itchka> Yes I think so
20:26:16 <itchka> In council we should consider those servers in Paris.
20:26:46 <itchka> as well
20:27:18 <blackcrack> oh, i should clean up this computermess, in the next minutes comes my replacement *bg* and the computer wheres be the monitoring be clean *giggle*
20:27:47 <blackcrack> oky, i mus go.. see you.. and mail :)
20:27:53 <blackcrack> byebye :)
20:28:01 <itchka> Bye Blacky
20:31:19 <tapwag> Just one question: Are these soyou-Servers physical or virtual machines?
20:32:22 <itchka> I'm pretty sure they are physical but I couldn't swear to it.
20:34:45 <tapwag> itchka, We might need to pay for remote hands if we want blacky's hard drives to be integrated
20:35:48 <tapwag> opening up the case and connecting the power and SATA cable
20:36:18 <itchka> tapwag: soyoustart offer certin add-ins to their servers I can't remember what they are at the moment but \i wouldn't mind betting that there are some storage options.
20:37:38 <itchka> or maybe we use the extra money to go up one level in the server option. It will have to be checked out.
20:37:42 <tapwag> itchka, I am pretty sure that there are storage options
20:39:14 <tapwag> itchka, Just one question: You are president but who is sorts of the "financial minister" an association needs to have?
20:39:39 <itchka> JCL is the treasurer.
20:39:57 <bero> tapwag: I'm pretty sure they aren't open to that because they want to rent their more expensive servers rather than allowing people to upgrade their cheapo servers
20:39:57 <tapwag> It is just I am a bit concerned on how much money we have in the bank
20:40:43 <tapwag> bero, Yes, they buy them from Dell and what have you and they want to get people renting their stuff
20:41:35 <tapwag> bero, And yes hosters very often buy a lot of rubbish
20:43:46 <itchka> I just checked and bero is right the only extra storage they offer is backup and that is quite expensive (but is fully managed)
20:47:16 <itchka> bero: Do we know if _TPG is coming?
20:47:42 <tapwag> itchka, I think we might to get autority on the physical devices themselves
20:48:34 <tapwag> But again this would be a question on our financial funds
20:48:45 <itchka> tapwag: How did you come to that conclusion?
20:49:59 <itchka> I can't see it on their site.
20:50:58 <tapwag> itchka, It comes from something that Jean Manuel said during Linuxtag 2014 which was about about "Datenhoheit". I do not know how to translate this into English, sorry.
20:51:12 <tapwag> Hoheit is something like authority
20:52:06 <tapwag> itchka, It is that I like the idea of having 2 of Blacky's hard drives popped in
20:52:55 <itchka> tapwag: When first conceived the new abf we looked at how people could contribute to Openmandriva the cheap servers from people like soyoustart and kimsufi were cheap enough for people to contribut by renting one to serve as an abf node.
20:54:32 <tapwag> In that case we would need to make a call to decicate computing power to OpenMandriva.
20:54:59 <tapwag> But I seriously doubt many people will jump on it
20:55:10 <itchka> It had the benefit that the individual was in control of the finance they simply rnted the server and made it available to abf. This greatly reduce the cost of running abf from what it was costing to support ROSA's abf.
20:55:11 <fedya> n3npq need your help
20:55:33 <n3npq> okay? what … ?
20:55:34 <fedya> n3npq i need to enable networking on docker0 interface
20:55:42 <itchka> tapwag: Yes that is the other part of the plan.
20:56:54 <itchka> If Blacky want to contribute the best think he can do is offer his basement server as a build node.
20:56:56 <tapwag> itchka, I do like the decentralised structure and there has been a lot of effort going into this.
20:57:39 <tapwag> itchka, But I also must admit that I am fascinated on what IBM has on offer.
20:58:39 <itchka> tapwag: The way abf is designed we can utilise both. Just as fedya has with Google server he has connected.
20:58:46 <tapwag> itchka, This would indeed lock us in with IBM, I appreciate that.
21:02:20 <blackcrack> rehy.. at home now.. i guess, we could use the ibm-offer as real turbo/booster if more computingpower need start the one from IBM and boost the performance of the ABF or some ..
21:06:13 <tapwag> blackcrack, itchka Just some of the thoughts I went into the conversations with IBM: We are getting short of computing power. As you have invented "Watson" supercomputer, would you mind sharing some resources with us?
21:07:31 <tapwag> I never mentioned the name of "OpenMandriva" I has just been like in the old geocities.com days where you would get webspace for free
21:08:15 <tapwag> It has just been
21:10:43 <tapwag> It's getting late klebedeff itchka Shall we close the meeting?
21:11:05 <klebedeff> i think yes tapwag
21:11:20 <klebedeff> itchka closing for today?
21:11:29 <tapwag> klebedeff, From my perspective: Please go ahead.
21:12:03 <bero> we have another topic, but we should really talk about that when TPG is there
21:12:09 <bero> so yes, close for today sounds good
21:12:29 <klebedeff> unless someone else has AoB this meeting will close in 5 minutes
21:12:39 <itchka> Ok by me
21:14:28 <blackcrack> #chwido Close Meeting ?
21:15:19 <blackcrack> chwido: bark around
21:15:19 <chwido> refuses to bark at such a nice soul.
21:15:45 <blackcrack> chwido: meet ?
21:16:12 <blackcrack> chwido: bite ChanServ
21:16:13 * chwido thinks about biting ChanServ but decides to not to.
21:17:17 <blackcrack> chwido: leak klebedeff
21:17:19 <blackcrack> *g*
21:17:45 <klebedeff> :)
21:17:52 <blackcrack> hihi
21:17:56 <klebedeff> #endmeeting