[13:10] <xnox> cyphermox, i'm pondering to enable disk-detect/multipath/enable to true by default on s390x, because it works fine there.
[13:11] <xnox> i also think it does work fine everywhere else too, and if things are multipath, we really shouldn't be presenting each path as a separate disk....
[13:11] <xnox> should we enable it by default to true on all architectures?
[13:11] <xnox> or shall i stick to s390x only for now?
[13:26] <cyphermox> xnox: why even on s390x?
[13:26] <cyphermox> you may have a not fully provisioned box without all the drives in, I suppose.
[13:27] <xnox> cyphermox, on s390x there are dasd drives - these are non-multipath ones, and s390-zfcp which is fibre-channel, almost always properly multipath. And I have a tonne of bug reports from IBM and customers, who are failing to discover diskt-detect/multipath/enable
[13:28] <xnox> and it just works, and one can even boot off multipath device on s390x.
[13:28] <xnox> so on s390x people expect multipath with zfcp.
[13:45] <cyphermox> heh
[13:46] <cyphermox> the same logic goes for ppc64el then; perhaps it would be best to make it the default there too
[13:47] <cyphermox> ie. the controller can multipath, then it's up to you whether you want to use it or not
[13:48] <cyphermox> OTOH, could you also not have zfcp?
[13:49] <cyphermox> if you test it to a reasonable level of certainty that it doesn't break the obvious test cases, I'm fine with enabling multipath anywhere / everywhere, really
[13:50] <cyphermox> normally multipath should work fine and not take over non-multipath drives
[13:50] <cyphermox> xnox: maybe ask rharper / smoser's opinions too
[13:50] <xnox> cyphermox, well, i tested it sufficiently on s390x
[13:50] <xnox> never touched ppc64el yet (as in physical machines, i was on kvm instances before)
[13:50] <xnox> not sure about other arches.
[13:58] <cyphermox> xnox: well, the same logic is valid on all arches -- you don't usually have multipath-capable things, but if you do, it's nice to have it enabled
[13:58] <cyphermox> it's always dependent on you doing the right hardware setup first
[13:59] <cyphermox> even with FC -- you don't necessarily put two paths to NAS
[14:01] <xnox> so shall i shoot a email out to enable it by default everywhere then? and solicit feedback?
[14:36] <cyphermox> would be best especially by now
[15:46] <cyphermox> xnox: how are you testing the translation magic?
[15:46] <cyphermox> it certainly looks to be retrieving the main-menu translations correctly here for russian
[15:46] <cyphermox> not that I can read it, but I get upside down and reversed characters
[15:46] <xnox> cyphermox, boot amd64 server iso in qemu-kvm, select russian in boot menu, start install, go back, main-menu is in english
[15:47] <cyphermox> that's exactly what I did
[15:47] <cyphermox> thing is, you need to go through localechooser first
[15:47] <xnox> let me sync in new server image.....
[15:47] <xnox> hm.
[15:48] <cyphermox> perhaps we should preseed more forcefully from gfxboot, and that would fix that part..
[15:48] <cyphermox> except if someone picks a lang on the gfxboot menu and then wants to use something different, it's going to be ugly
[15:50] <xnox> huh, i went through locale chooser changed russian to latvian and things work fine, e.g. everything is in latvian now.
[16:04] <CarlFK> d-i partman-auto/disk string /dev/sda - fails on a box that gives sda to a card reader, and there is no card, so it errors.     Is there a way to say "find a drive and use it"?  (which some would call dangerous, but I am ok with that)
[16:11] <cyphermox> CarlFK: not aside from using an early_command to write your own script to do that
[16:13] <CarlFK> bleck.  too much work/clutter for this problem
[16:13] <CarlFK> I can pass it in as a kernel parameter, right?
[16:37] <xnox> cyphermox, i wonder if translation menus work on e.g. powerpc.
[16:37] <xnox> infinity, is main menu translated on powerpc?
[16:37] <cyphermox> why would it be different there?
[16:37] <xnox> big endian like s390x =)
[16:37] <cyphermox> I can try it just as soon as this meeting is done
[16:38] <cyphermox> oh, actually, before since I can use another system to avoid cutting the link