[10:17] <caribou> apw: I owe you an apology : I handed you a slightly rotten patch for initramfs-tools :-/
[10:17] <apw> caribou, heh ooops
[10:17] <apw> that might explain cking's failure then
[10:18] <caribou> apw: yes it does; the patch was done on trusty and  one variable name changed
[10:18] <apw> caribou, np, it went into my test PPA for this reason
[10:18] <cking> and i like testing things :-)
[10:18] <caribou> cking: :)
[10:20] <caribou> apw: the variable ${root} got changed by ${dev_node} b/w the two versions
[10:21] <apw> caribou, wanna mail me an update and i'll respin it
[10:21] <caribou> apw: you should even be able to edit the quilt patch manually; I'll email you the details
[10:22] <apw> ok, its not a quilt patch any more because we don't do that on native things and its in git and ...
[10:22] <apw> but details of whatall exactly to change is good
[10:24] <caribou> apw: true, that's a native package
[10:24] <caribou> apw: ok, I just sent the mail
[10:24] <caribou> apw: just wave at me when you want me to re-test it
[10:30] <apw> thanks
[13:29] <xnox> apw, i've upgraded to linux-generic 4.3.0.6.7 package from proposed, yet my /boot/vmlinuz and /boot/initrd.img symlinks still point at the old -5-generic abi files
[13:29] <xnox> how should i go about debugging this? which bit is responsible to update them?
[13:30] <apw> xnox, and you got linux-iamge etc installed ok ?
[13:30] <apw> xnox, as they are updated from postinst there
[13:30] <apw> xnox, the same postinst which builds you initramfs
[13:30] <xnox> yeah i have linux-image-generic                  4.3.0.6.7
[13:31] <xnox> let me read them, and re-run them again. (aka downgrade make sure everything is clean and check what's going on)
[13:31] <apw> no that is the meta package, i mean the real binaries
[13:31] <apw> linux-image-xxxxx-6-generic is the one which installs the links
[13:31] <xnox> i have linux-image-4.3.0-2|5|6-generic, and headers, and extra
[13:31] <apw> and the links will always be the _last_ one you installed
[13:31] <apw> not the highest
[13:32] <apw> and of course they are ignored for grub, not sure about zipl
[13:32] <apw> do we generate a menu using the real files or just "current and previous" using the links ?
[13:32] <xnox> at the moment we don't generate a menu, no.
[13:33] <apw> so i assume it is static pointing at the links
[13:33] <xnox> i do want to start generating menus, for all variants.
[13:33] <xnox> yeap.
[13:34] <apw> the links are right on my local machine for whatever that is worth
[13:34] <apw> not that i would notice if half the time they were not
[13:35] <xnox> apw, the whole lot is a mess. also is it even allowed to have perl as postinst scripts?!
[13:35] <apw> xnox, heh, we've done it for years, so possibly
[13:39] <xnox> so i download -5- of headers, headers-generic, image, image-extra.
[13:39] <xnox> i will fully uninstall and purge -5-
[13:39] <xnox> install these .debs again
[13:40] <xnox> and expect the symlinks to be update.
[13:40] <xnox> and expect the symlinks to be updated.
[13:40] <xnox> e.g. when -5- gone, symlinks should point to -6-
[13:40] <xnox> and when -5- is installed again, symlinks should point to -5-
[13:42] <apw> should point to whatever prev pointed to
[13:42] <xnox> removing linux-image-4.3.0-5-generic failed.
[13:42] <xnox> symlinks still point to -5- and running postinst.d/zz-zipl fails cause the config points at broken symlinks
[13:42]  * xnox files a bug.
[13:45] <xnox> apw, http://paste.ubuntu.com/14495976/
[13:48] <xnox> apw, i wonder if /etc/kernel/postrm.d/zz-zipl is run too early.
[13:49] <xnox> reinstalling -6 on top, doesn't fix the symlinks, as
[13:49] <xnox> Not updating initrd symbolic links since we are being updated/reinstalled 
[13:49] <xnox> (4.3.0-6.17 was configured last, according to dpkg)
[13:50] <xnox> which is true, but when one has broken symlinks on disk, surely they should be fixed to point at something real.
[13:54] <xnox> apw, anything wrong in this case - http://paste.ubuntu.com/14496011/ ?
[13:58] <xnox> apw, i am an idiot
[14:00] <xnox> ...
[14:00] <xnox> aghrrr
[14:01] <xnox> so kernel updates / symlinks, and zipl-installer is using symlinks in /boot/
[14:01] <xnox> i'm not sure what creates symlinks in /boot/ or manages them.
[14:01] <xnox> also not sure if zipl can handle symlinks in /, rather than /boot/
[14:04]  * xnox ponders which way it's better to fix these.... to follow kernel lead, or to follow zipl-installer lead
[14:05] <xnox> i guess the real answer is, that i should generate entries for things in /boot/ properly.
[14:05] <xnox> and zipl-installer expects to do do_bootloader by the kernel too, and we have no zipl support there....
[14:31] <xnox> apw, is it fair to say that perl postinst scripts in linux kernel, are out of date with respect to changes done in e.g. debian? and it looks like the kernel clobbers and writes out the wrong kind of kernel-img.conf, unlike what base-installer / live-installer should be writting.
[14:31] <xnox> and e.g. in debian linux maintainer scripts never create kernel-img.conf
[14:35] <apw> xnox, they are very idffernet to debian yes, based on an older one from there yes
[15:14] <xnox> apw, i think my machine has a botched installation on it. and e.g. fresh cloud images have 'link_in_boot = Yes' in the /etc/kernel-img.conf and then everything works correctly. 
[23:57] <phillw> Hi folks, with all the rumours going around, is 16.04 LTS going to be 4.3 or may it be 4.4, or 4.5 ?