20:02:39 <DavidWHodgins> #startmeeting
20:02:39 <Inigo_Montoya> Meeting started Thu Mar  8 20:02:39 2018 UTC.  The chair is DavidWHodgins. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:02:39 <Inigo_Montoya> Useful Commands: #action #agreed #help #info #idea #link #topic.
20:02:44 <DavidWHodgins> #topic * Who's new? - If you are then come and say hello.
20:03:42 <tarazed> Lewis sends his apologies - will be away for a week but hopes to get to the meeting next week.
20:04:27 <bri_an> tell him we wish him well
20:04:37 <DavidWHodgins> Don't see any new names in the list. As anonymizer domains such as bitmessage.ch are blocked as we find out about them, I doubt we'll see zfic6 here
20:05:00 <bri_an> I wondered
20:05:19 <DavidWHodgins> #topic Meltdown/Spectre updates status
20:05:30 <wilcal> More kernels?
20:05:40 <DavidWHodgins> tmb: Still expecting more microcode updates?
20:06:35 <tmb> DavidWHodgins, yeah, Intel have released microcode from haswell and up, but only to hw vendors... not to public
20:06:47 <DavidWHodgins> So more waiting
20:07:05 <tmb> yeah...
20:07:17 <DavidWHodgins> #topic Testing updates - status and questions
20:07:54 <wilcal> I went through the Qt files in a client and that seemed to go well
20:07:55 <DavidWHodgins> Taking note of the problems Herman is seeing, anyone else seeing similar problems with the qt/kf5 stack updates?
20:08:10 <tarazed> We should acknowledge the return of MrsB to advisories and testing.  Thanks Claire.
20:08:12 <tjandrews> Once fully installed, Plasma 5.12.2 has been working great for me.
20:08:57 <tarazed> Lewis thinks that there are still problems though.
20:09:06 <Bequimao> the same here.
20:09:13 <tarazed> I am not taking part in this discussion but have some quotes from him.
20:09:22 <neoser10> The problema are detected in real hardware???
20:09:43 <DavidWHodgins> I'm wondering if due to trying to install updates before they were all ready, some rpm packages got removed that shouldn't have been
20:09:43 <neoser10> Problems,sorry multi language correction
20:10:04 <tarazed> Lewis was also wondering if running the update stacks in a particular order would be more effective, so that they can be pushed one at a time, like Qt5, KF5 or LXQt, then KDE applications?
20:10:32 <tmb> no, that wont work
20:10:58 <DavidWHodgins> They are tied together. kf5 and lxqt both need the qt update, and the qt update by itself would break both kde and lxqt
20:11:07 <tjandrews> I have been concerned that we'd run into problems due to mirror-syncing with such a massive update.
20:11:13 <Bequimao> there are some issues resolved that I did not file, because I was waiting for the Plasma updates-
20:11:14 <tmb> if you push qt first, current plasma will fall apart... and since the rest depends on new qt...
20:11:52 <wilcal> My heads spinning
20:12:13 <DavidWHodgins> We will, but as there is no way to stop mirrors from syncing in the middle of the packages being moved from testing to updates, no way to completely avoid it.
20:12:44 <DavidWHodgins> As different mirrors sync at different times, there is no "good time" to do it.
20:13:14 <Bequimao> The painel broke. Best is to invoke a full upgrade in a virtual console.
20:13:39 <tmb> I think I will try a different "more manual" push with this update..
20:14:24 <tmb> basically, push the updated packages, but no new hdlists... then wait ~24 h and push the hdlists
20:14:39 <neoser10> That is a good idea
20:14:41 <DavidWHodgins> That might work
20:14:53 <wilcal> "might" oh my
20:15:02 <DavidWHodgins> :-)
20:15:19 <DavidWHodgins> Don't think we've used that technique before, so I don't know for sure.
20:15:35 <tjandrews> Some way so that mirrors aren't syncing on a partial list.
20:15:47 <Luigi12_work> I think people having problems are using NVIDIA cards, but NVIDIA cards are always problems
20:16:27 <wilcal> can i just update plasma-desktop and all the packages fall in line?
20:16:43 <DavidWHodgins> For kde, gnome, and similar, all we can fix are packaging problems. Individual program problems have to be referred upstream.
20:16:48 <tjandrews> None here, but I've only used my nvidia machine with it for about an hour so far.
20:16:53 <tmb> well, it should make the update more stable as mirrors have 24h to catch up before the new hdlists comess
20:16:57 <Bequimao> Well I don't remember well what was the need of qt5/kf5/plasma updates. We discussed it before the release of Mageia 6.
20:17:34 <DavidWHodgins> A massive number of bugfixes are included in those updates
20:18:02 <tmb> well, basically current plasma in mga6 suck... so the only way to get something "useful" is a huge update like this
20:18:05 <DavidWHodgins> Anyone trying to report a problem will be told to update to the current version
20:18:14 <DavidWHodgins> As in report to kde
20:18:25 <wilcal> I still see lots of wrinkles with Plasma
20:18:34 <wilcal> some of them serious
20:18:49 <DavidWHodgins> Yes, but less than before, from what I've seen
20:18:53 <tmb> wilcal, its called "progress" :)
20:20:05 <bri_an> hmm - anybody tried it with 32bit yet?
20:20:11 <DavidWHodgins> I still need to add the advisories to svn. I'll be going through the list of rpm packages on the mirrors, comparing it to the bug reports to make sure none are missed. That will take a day or so.
20:20:22 <DavidWHodgins> bri_an: Yes, but only under vb so far.
20:20:48 <bri_an> ok - I may try a build over the weekend and see what happens on my plasma laptop
20:21:17 <tjandrews> I may give it a shot on 32-bit later today. I have a 32-bit install on my laptop.
20:21:17 <bri_an> I'm about to rebuild it anyhow
20:21:43 <DavidWHodgins> Unless we find a specific packaging problem that needs to be fixed, I'd like to plan on validating the updates Monday.
20:22:00 <DavidWHodgins> Don't want to keep them too long as they are blocking some security updates.
20:22:02 <tmb> also for 32bit... preferably on some "old" hw that dont have sse2 (to see if QT stack is still working as intended
20:22:20 <bri_an> tmb:  I doubt that could run plasma
20:22:20 <DavidWHodgins> Ouch. That would be slow.
20:22:36 <bri_an> does mate need qt5?
20:22:55 <bri_an> sorry my ignorance on that
20:23:04 <tarazed> Depends on what applications you run.
20:23:08 <tmb> you forget there is also a new lxqt rolling out... its the QT part that is important... I could not care less about plasma
20:23:32 <tarazed> +1
20:23:38 <Bequimao> Someone marked Application Stack Update as Ok. I would not dare.
20:23:49 <bri_an> hmm - so maybe build a lxqt and then attack with with qt5.9 and watch it
20:24:32 <bri_an> ?
20:25:01 <DavidWHodgins> You mean install lxqt, then the update and test it?
20:25:18 <bri_an> freshbuild of MGA6 using lxqt and test it.
20:25:42 <bri_an> it's a 64-bit sempron, so probably has sse2
20:25:46 <DavidWHodgins> Certainly doesn't hurt. The more testing the better, but try to be done testing by Monday.
20:26:01 <bri_an> I'll hit the laptop then.
20:26:21 <bri_an> sempron on 754 fyi
20:27:30 <tmb> bri_an, that one has sse2
20:27:48 <bri_an> :-(  sorry, it's the oldest I have.  I call it my "dog with fleas"
20:27:53 <tarazed> bri_an: just ran a query on taks-mate and found one instance of qt - pavucontrol-qt
20:27:56 <hviaene> Has pentium M?
20:28:20 <bri_an> Looking that up.
20:28:21 <DavidWHodgins> I'm not sure
20:28:41 <tmb> Pentium M should have sse2
20:29:10 <DavidWHodgins> This 10 year old laptop has sse, sse2 and ssse3
20:29:26 <tmb> I think one should be able to test with kvm as that should be able to hide sse+
20:29:43 <tmb> qemu I mean
20:30:49 <DavidWHodgins> I'm so used to vb I only use qemu when testing an update for it or xen
20:30:58 <bri_an> xen/qemu I haven't tried that since going to grub2.  That would make me think hard
20:31:09 <DavidWHodgins> :-)
20:31:33 <DavidWHodgins> That's partly why I'm still using grub legacy on most of my installs
20:31:47 <Bequimao> kvm/qemu works fine here. why test it on old hardware?
20:32:39 <DavidWHodgins> Testing it with sse2 hidden should show if it will work on older hardware (like it's supposed to), without having to actually use old hardware.
20:32:53 <neoser10> Pentium 3 and pre pentium 4 lacks the sse2
20:33:30 <DavidWHodgins> Thanks
20:33:41 <tmb> and athlon 64 / XP and older... and older celerons
20:34:23 <DavidWHodgins> IIRC my old celeron system that died 2 years ago had sse2
20:34:56 <bri_an> gave away my mp1900 x2 box
20:34:58 <DavidWHodgins> Do arm systems suport sse2?
20:35:08 <wilcal> My celeron stuff found the recycler before I moved
20:35:36 <tmb> yeah.. it is getting harder to find.. and harder to work around...
20:36:12 <DavidWHodgins> I'll skip the other individual updates, as they are all packages we've had before
20:36:23 <DavidWHodgins> #topic Team elections
20:36:33 <bri_an> oh that time again
20:36:35 <DavidWHodgins> It's that time of year again.
20:36:37 <DavidWHodgins> :-)
20:36:47 <wilcal> That's right we're back in elections time again
20:37:17 <DavidWHodgins> I'll send a msg to the ml. Would like to have all nominations in by the next qa meeting
20:37:22 <wilcal> we should list who is officially elected to what right now
20:37:25 <neoser10> The processors date is 1999 that date was the date to SIMD extension
20:37:28 <DavidWHodgins> Will do
20:37:55 <wilcal> cuz we all kinda do everything at some time :-)
20:38:17 <DavidWHodgins> I'm surprised I haven't seen other teams start elections yet.
20:38:18 <bri_an> let me know who wants to be elected
20:38:48 <DavidWHodgins> Guessing that since Mageia wasn't represented at Fosdem this year, it's been kinda forgotten
20:39:27 <DavidWHodgins> I'll post a reminder to the council ml too
20:39:27 <wilcal> that's 2-years in a row not at Fosdem
20:40:09 <DavidWHodgins> #topic * Anything else?
20:40:25 <bri_an> not from me
20:40:26 <DavidWHodgins> So, anyone have any other topics or questions?
20:40:27 <wilcal> not from me
20:40:29 <tmb> speaking on kernels ... I will submitting kernel 4.14.25 to testing (most likely tomorrow) with security fixes, followup fixes to spectre/meltdown, and some more hw support/fixes...
20:40:49 <wilcal> there goes my weekend :-0
20:40:50 <bri_an> ok - will look for those tomorrow
20:40:58 <DavidWHodgins> At least they don't require the qt update. :-)
20:40:58 <neoser10> Is already the vboxaddition forma that krnl
20:41:34 <neoser10> for that kernel???
20:41:37 <tjandrews> But tests on machines that already have the QT update are need too, no?
20:41:50 <wilcal> kernel testing has become a piece of cake against all the other stuff
20:41:53 <DavidWHodgins> Shouldn't matter.
20:41:59 <rindolf> tmb: https://www.kernel.org/
20:42:00 <[mbot> [ The Linux Kernel Archives ]
20:42:15 <rindolf> tmb: there is only 4.14.24
20:42:35 <DavidWHodgins> I'm having problems with my i586 test installs, though
20:42:57 <tmb> rindolf, 4.14.25 is in review right now... and current 4.14.24-3 in cauldron have all bits of it :)
20:43:08 <rindolf> tmb: ah
20:43:09 <DavidWHodgins> The i586 installs are acting as if the partitions are not aligned
20:43:35 <bri_an> now that sounds strange
20:43:47 <tmb> DavidWHodgins, where/what/how do you see that ?
20:44:02 <DavidWHodgins> On my "green" spinning rust drive with a 512 logical 4k physical sector size I'm seeing massive io/waits during any large disk write
20:44:18 <DavidWHodgins> The sectors for the partition starts are aligned
20:44:38 <tmb> DavidWHodgins, are you running 4.14.20 on them ?
20:44:39 <bri_an> does it clear on reboot and run well fro awhile?
20:44:53 <DavidWHodgins> I tried creating a copy of the Mageia 6 i686 installs on one of my ssd drives, and it's showing the same problem.
20:45:10 <DavidWHodgins> bri_an: If I wait it's ok, it just really slows down testing.
20:45:43 <DavidWHodgins> tmb: It's been going on for several kernel updates now. Not sure when it started. Been trying to make sense of it before bringing it up
20:46:05 <DavidWHodgins> Only affects the i686 installs. The x86_64 installs on the same drives are fine
20:46:32 <tmb> well, there is also a bug in current 4.14.20 that causes scsi layer lockups at times, something I've fixed in later kernels
20:47:02 <DavidWHodgins> I'll see if tomorrow's kernel update fixes it. Fingers crossed!
20:47:16 <neoser10> I have all the kernels from 4.9 series and 4.14.20 i will review the journald
20:47:26 <bri_an> DavidWHodgins:  what do you call big file?
20:47:46 <tmb> also you caould try to change io scheduler if it helps / changes anything
20:47:46 <DavidWHodgins> Installing 5 kernels in one urpmi command
20:48:18 <bri_an> ok
20:48:18 <DavidWHodgins> bri_an: So it's 5 kernels, plus 5 mkinirds
20:48:27 <neoser10> The ISO copied between devices
20:48:49 <bri_an> ok was curious.  I was just moving around about 1 GB of data, but didn't see much impact on my i586 AMD
20:49:03 <tjandrews> And any dkms-related building.
20:49:32 <DavidWHodgins> There is dkms, but what I'm seeing are device waits with low cpu usage
20:50:11 <DavidWHodgins> Even on the ssd drive, which the x86_64 installs are ok on
20:50:58 <DavidWHodgins> It could be the i686 driver for the controller rather then an alignment or drive problem
20:50:58 <tmb> DavidWHodgins, what hw/chipset ? ssds should not take a big hit
20:51:58 <DavidWHodgins> JMB362 Jmicron sata controller
20:52:22 <tmb> ah... you are screwed then...
20:52:32 <DavidWHodgins> Awww!
20:52:35 <tmb> iirc that one is known to have issues...
20:53:11 <DavidWHodgins> Weird that it only impacts 32 bit installs, and only showed up a while ago
20:54:15 <DavidWHodgins> Time to start putting together a new testing system soon :-)
20:54:57 <tmb> yeah, it usually comes when subsystem gets more and more optimizations... the controller cant keep up... and since most upstream prefers 64bit theese days ... that part will work better
20:54:58 <DavidWHodgins> Countdown time I guess
20:55:00 <wilcal> Something with enough Dram to build a little Vbox internal network. About 16GB
20:55:18 <wilcal> Thanks for being in charge David
20:55:33 <DavidWHodgins> I have 16GB ram in that system
20:56:05 <DavidWHodgins> No firmware updates available for it (I've been watching for spectre/meltdown fixes).
20:56:28 <DavidWHodgins> Likely too old now (6 years) to get any.
20:56:33 <DavidWHodgins> T - 5
20:56:36 <DavidWHodgins> 4
20:56:40 <wilcal> bye all
20:56:40 <DavidWHodgins> 3
20:56:43 <DavidWHodgins> 2
20:56:46 <DavidWHodgins> 1
20:56:52 <DavidWHodgins> Thanks for coming everyone!
20:57:00 <DavidWHodgins> #endmeeting