IRC logs for #baserock for Monday, 2015-02-16

*** petefoth has joined #baserock07:28
*** zoli__ has joined #baserock08:02
*** sherm_ has joined #baserock08:43
*** rdale_ has joined #baserock08:43
*** mariaderidder has joined #baserock08:48
*** sambishop has joined #baserock08:53
*** zoli__ has quit IRC09:03
*** zoli__ has joined #baserock09:03
*** bashrc has joined #baserock09:07
*** jonathanmaw has joined #baserock09:18
*** tlsa_ is now known as tlsa09:29
*** tlsa_ has joined #baserock09:31
*** tlsa has quit IRC09:33
*** tlsa_ is now known as tlsa09:33
pedroalvarezHi baserockers! I've been researching about how to "kill" /[s]bin /lib[64] (https://storyboard.baserock.org/#!/story/11) And I've found that if a chunk is being unpacked in the staging area, the chunk things in /bin , and the staging area has already a symlink from  /bin to /usr/bin. then unpacking will success and things in /bin will go to /usr/bin.09:38
pedroalvarezThis behaviour makes me wonder if it's enough with modifying build-essential.morph to make it produce the symlinks to /usr, or we want to fix all the chunks that are using the non /usr dirs09:39
*** gary_perkins has joined #baserock09:39
pedroalvarezThe former will work  with our current staging area construction, but maybe not if we change it09:40
tlsaare there many chunks that use the non /usr locations?09:41
pedroalvarezin a devel system not too many. around 7, but some of them don't support installing to /usr in their scripts09:43
*** CTtpollard has joined #baserock09:44
*** edcragg has joined #baserock09:54
*** ssam2 has joined #baserock10:00
*** ChanServ sets mode: +v ssam210:00
richard_mawwe should make /bin, /sbin, /lib, /lib32 and /lib64 symlinks to their /usr equivalents, since some stuff in systemd works better, such as ProtectSystem10:08
richard_mawProtectSystem makes /usr read-only, but has no provisions for /bin, /sbin etc by default10:08
ssam2we should do that indeed10:08
ssam2we could remove some hacks to override busybox versions of the tools, as well10:08
richard_mawso we'd either have to move everything to /usr, or patch systemd's ProtectSystem to protect more directories by default10:09
*** paulwaters_ has quit IRC10:09
richard_mawthere's also the ephemeral containers stuff, which can work by bind-mounting /usr read-only into a tmpfs10:09
radiofreefedora uses symlinks10:16
pedroalvarezand so I did in my WIP branch10:16
*** paulwaters_ has joined #baserock10:23
*** Krin has joined #baserock10:26
*** paulwaters_ has quit IRC10:27
*** paulw has joined #baserock10:27
richard_mawsymlinks is the most sane approach10:27
pedroalvarezas a side note: don't ever ever link to /usr/bin from /bin in stage2 chunks10:28
richard_mawsystemd will probably make ths symlinks for you these days if you don't have them, but we need them to be available at build time for various compatibility reasons10:28
radiofreessam2: pedroalvarez: i'm going to hold off on that revert commit, it at least seems to build on armv8...10:47
pedroalvarezradiofree: okay10:48
*** jonathanmaw has quit IRC11:04
*** jonathanmaw has joined #baserock11:19
*** CTtpollard has quit IRC11:33
*** CTtpollard has joined #baserock11:36
radiofree"ERROR: The ssh-rsync write is for upgrading existing remote Baserock machines. It cannot be used for an initial deployment."11:52
radiofreeoop11:52
radiofrees11:52
radiofreehm11:52
radiofrees/hm/nm11:52
ssam2use 'morph upgrade' not 'morph deploy' for that11:57
radiofreeyep!11:57
pedroalvarezI'm  mounting a rawdisk image, putting a file inside, booting the image, and not seeing the file. I'm mounting againg the image, and the file is there. What am I missing?12:08
ssam2sync ?12:10
ssam2not sure12:10
radiofreepedroalvarez: where did you put the file12:19
radiofreesync is also a good idea12:19
pedroalvarezradiofree: root folder of the run subvolume  of the factory system12:19
radiofreepedroalvarez: and that's the only one on there?12:20
pedroalvarezyeah :)12:21
pedroalvarezSam pointed me to the possible cause: different versions of btrfs in the host I'm mounting the image to inject the file, and in baserock12:22
radiofree:/12:22
pedroalvarezhold on, the file is there, but it wasn't showing up in /root12:23
pedroalvarezbut it is when I mount the /dev/sda disk and look inside12:23
pedroalvarezok12:24
pedroalvarezeverything is ok12:24
pedroalvarez\/mount is a state folder, baserock mounts the state/root subvolume in /root. That's why I wasn't seeing the file12:27
*** wschaller has joined #baserock12:40
radiofreeok, found the problem13:00
radiofreenew m4 checks for a .git folder, if found, it enables all the gcc warnings13:01
radiofreethe default for a release/tarball should be --disable-gcc-warnings, will resubmit soon13:01
pedroalvarezradiofree: great!13:02
tlsaif I have a chunk in two stratum (xserver in x-xorg and x-wayland) should the chunk have different names in each, or the same name?13:53
*** mariaderidder has quit IRC13:53
*** mariaderidder has joined #baserock13:54
ssam2tlsa: different names in each13:54
tlsaok13:55
ssam2use @, e.g. xserver@x11 and xserver@wayland13:55
tlsaright13:55
* tlsa is a bit unsure of the graphics stack stuff. Could all the stuff like the various fonts related chunks be moved out of x-wayland/x-org (previously x-generic) and into x-common?13:57
tlsadoes anything in the build-depends: of another chunk in the statum need to have the @...14:00
tlsa?14:00
franredI put back the patch which creates an openstack bsp but I think we should add the following patch http://paste.baserock.org/cuyixoseye, could someone do a fast review to it?14:02
CTtpollardhi following this http://wiki.baserock.org/guides/deploy-trove-to-openstack/ , in the config.sh where <your public key here> do you place your ssh key as a single string, and without or without the <> ?14:19
pedroalvarezCTtpollard: yup14:21
pedroalvarezCTtpollard: if you are wondering why I'm not answering your second question. I really answered it :)14:21
CTtpollarderm14:22
CTtpollardok14:22
CTtpollard*with or without14:22
pedroalvarezwithout I believe14:23
pedroalvarezCTtpollard: yes, without14:24
CTtpollardty pedroalvarez14:33
ssam2tlsa: yes, @ is part of the name so it'll need to be there14:37
tlsaok, so @ is convention, not something meaningful to morph14:37
ssam2yeah14:37
ssam2re. fonts, probably makes sense to put them into x-common... would be good to ask jjardon when he's around, he's most active in this area14:37
tlsaI'd just made an x-fonts for those chunks, but I seem to have introduced a deps cycle (well, its hissting recursion limit jumping between get_cache_id and _calculate, and I guess that's what's happened)14:39
tlsahissting = hitting14:40
ssam2yeah, that means a dep cycle, nobody added a nice handler for that yet14:42
tlsadon't think I can move the fonts stuff to x-common without making x-common depend on graphics-common which is where freetype is14:42
tlsaalthough graphics-common depends on x-common, so maybe the x fonts stuff can go there14:43
tlsait's all a bit difficult to keep track of14:43
rdale_i've lost track of what is going on with xserver support too14:44
tlsaatm all the systems which used xserver without wayland are broken14:44
rdale_yes14:45
*** zoli__ has quit IRC14:47
ssam2tlsa: is it caused by http://listmaster.pepperfish.net/pipermail/baserock-dev-baserock.org/2015-January/010960.html ?14:52
ssam2and would reverting that patch fix it?14:52
ssam2or is it more complex ?14:52
tlsassam2: possibly, i'd need to test.  I was trying to make separate x-wayland and x-xorg stratum so that the other stuff that uses them can be as "minimal" as possible though14:54
ssam2that seems a reasonable approach, if it's possible. but the current definitions format can be quite limiting when trying to do stuff like that14:59
tlsayeah, ok15:00
*** wschaller has quit IRC15:05
pedroalvarezfranred: ok, I found the relevant commit for your patch ( http://paste.baserock.org/cuyixoseye )15:07
pedroalvarezhttp://git.baserock.org/cgi-bin/cgit.cgi/delta/linux.git/commit/?id=8993cf8edf42527119186b558766539243b791a515:07
franredpedroalvarez,  is that a +1 to my patch in our current BSPs? (I think I will have to extend to the rest of BSPs)15:10
pedroalvarezfranred: is a "I'm not sure that the patch is right"15:14
franredpedroalvarez, my patch was related with https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=76569115:18
mauricemoss_I've prepared a fix for the Morph MIPS patch and ran './check' with 27 failed scenarios (Unknown architecture x86_64l in morphology string), are they relevant? http://paste.baserock.org/owufafogas15:22
*** wschaller has joined #baserock15:26
pedroalvarezmauricemoss_: can I see what you did to detect the endianess?15:32
pedroalvarezISTM that you are also appending 'l' or 'b' for x8615:33
mauricemoss_pedroalvarez, https://github.com/mossmaurice/morph/commit/711ccad4b10c5f7e6530cc23c82ec47c0a0131bf15:34
mauricemoss_I'll check it15:34
pedroalvarezmauricemoss_: odd, that looks ok to me15:40
*** zoli__ has joined #baserock15:43
pedroalvarezmauricemoss_: so `morph print-architecture`  should print x86_64l15:43
pedroalvarez(the tests are using that command to get the architecture to generate test systems to build )15:44
mauricemoss_pedroalvarez, thanks! it prints 'x86_64'. I'll rerun `./check` to be sure, my VM got killed several times when running it. Have to assign more memory first.15:47
pedroalvarezmauricemoss_: is this because you are filling /tmp? I believe you can use another folder with more space15:48
pedroalvarezIIRC: `TMPDIR=/src/tmp ./check`15:48
mauricemoss_Not only, I remounted /tmp with 10gb, but my qemu had only 2gb ram.15:50
pedroalvarezmauricemoss_: then attach a second disk, mount it wherever, and use it in TMPDIR15:50
pedroalvarezmauricemoss_: anyway, I'm running the tests for your changes15:51
mauricemoss_ok, ta!15:53
*** wschaller has quit IRC15:54
pedroalvarezmauricemoss_: they passed15:55
mauricemoss_nice, cheers15:56
*** wschaller has joined #baserock15:59
*** zoli__ has quit IRC16:56
*** lachlanmackenzie has joined #baserock17:03
*** zoli__ has joined #baserock17:12
*** gary_perkins has quit IRC17:17
radiofreehttp://mason-x86-64.baserock.org is red :\17:19
radiofree Building failed for m4-tarball-misc17:21
radiofreeas far as i can tell from the log it went ok?17:21
pedroalvarezradiofree: yes, you are right, it builds, something else if failing17:21
ssam2disk was full on cache.baserock.org17:21
ssam2i deleted the 10000 oldest artifacts to try and free some space17:21
radiofreeah, so it's currently building?17:22
*** zoli__ has joined #baserock17:24
*** rdale_ has quit IRC17:25
*** pdar has quit IRC17:25
*** DavePage_ has quit IRC17:25
*** Zara has quit IRC17:25
*** robtaylor has quit IRC17:25
pedroalvarezradiofree: yeah, I guess so, given that it hasn't failed in the last 40 minutes17:27
radiofreethat's good, i thought i'd broken everything17:29
*** rdale_ has joined #baserock17:31
*** DavePage_ has joined #baserock17:31
*** pdar has joined #baserock17:31
*** Zara has joined #baserock17:31
*** robtaylor has joined #baserock17:31
*** zoli__ has quit IRC17:32
*** zoli__ has joined #baserock17:32
*** jonathanmaw has quit IRC17:36
*** zoli__ has quit IRC17:40
*** mariaderidder has quit IRC17:44
*** ssam2 has quit IRC17:44
*** bashrc has quit IRC17:52
*** rdale_ has quit IRC18:03
*** sherm_ has quit IRC18:10
*** Krin has quit IRC18:10
*** edcragg has quit IRC18:16
*** lachlanmackenzie has quit IRC18:18
*** wschaller has quit IRC18:37
*** zoli__ has joined #baserock18:56
*** zoli__ has quit IRC22:46
*** zoli__ has joined #baserock22:46
*** zoli__ has joined #baserock22:47

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!