openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Move ppc block-device default to right $ARCH https://review.openstack.org/471170 | 00:01 |
---|---|---|
*** farosas has quit IRC | 00:06 | |
*** jamielennox is now known as jamielennox|away | 01:17 | |
mmedvede | ianw: ack | 01:29 |
ianw | mmedvede: unfortunately ... i've pulled a thread and now the whole thing is untangling ... | 01:30 |
mmedvede | this is what happens when you refactor :) | 01:30 |
mmedvede | rfolco (not in the channel) was looking at the original prep partition ppc fix, but I believe that things have moved very far since it was last touched | 01:31 |
*** pmannidi has quit IRC | 01:33 | |
ianw | i'd be interested if they want to look again, especially in a few days, as i'd really like feedback from fresh eyes on how it is working | 01:33 |
mmedvede | at least ppc test is still failing correctly | 01:33 |
*** jamielennox|away is now known as jamielennox | 01:34 | |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Use picked nodes for later cmd_* calls https://review.openstack.org/469330 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Don't make image & loopdev functions static https://review.openstack.org/470845 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Move global mount tracking into state https://review.openstack.org/469752 | 02:20 |
*** pmannidi has joined #openstack-dib | 02:20 | |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Update tracing in block_device_create_config_file https://review.openstack.org/471163 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Move rollback into NodeBase object https://review.openstack.org/464105 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Use global state to check for duplicate fs labels https://review.openstack.org/469751 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Remove 'state' argument from later cmd_* calls https://review.openstack.org/469750 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Pad state dump https://review.openstack.org/471162 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Add state to NodeBase class https://review.openstack.org/469749 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Move ppc block-device default to right $ARCH https://review.openstack.org/471170 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Start at block-device mounting/unmounting cleanup https://review.openstack.org/471183 | 02:20 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Start at block-device mounting/unmounting cleanup https://review.openstack.org/471183 | 02:30 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Use picked nodes for later cmd_* calls https://review.openstack.org/469330 | 02:51 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Start at block-device mounting/unmounting cleanup https://review.openstack.org/471183 | 02:51 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Don't make image & loopdev functions static https://review.openstack.org/470845 | 02:51 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Move global mount tracking into state https://review.openstack.org/469752 | 02:52 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Update tracing in block_device_create_config_file https://review.openstack.org/471163 | 02:52 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Move rollback into NodeBase object https://review.openstack.org/464105 | 02:52 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Use global state to check for duplicate fs labels https://review.openstack.org/469751 | 02:52 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Remove 'state' argument from later cmd_* calls https://review.openstack.org/469750 | 02:52 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Pad state dump https://review.openstack.org/471162 | 02:52 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Add state to NodeBase class https://review.openstack.org/469749 | 02:52 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Move ppc block-device default to right $ARCH https://review.openstack.org/471170 | 02:52 |
*** chhavi has joined #openstack-dib | 03:30 | |
*** hashar has joined #openstack-dib | 07:21 | |
*** hashar_ has joined #openstack-dib | 07:30 | |
*** hashar has quit IRC | 07:33 | |
*** hashar_ is now known as hashar | 07:51 | |
openstackgerrit | Merged openstack/diskimage-builder master: Pad state dump https://review.openstack.org/471162 | 08:21 |
*** pmannidi has quit IRC | 09:25 | |
*** jamielennox is now known as jamielennox|away | 10:10 | |
*** jamielennox|away is now known as jamielennox | 10:16 | |
*** brault has joined #openstack-dib | 11:10 | |
*** brault has quit IRC | 11:26 | |
*** brault has joined #openstack-dib | 11:48 | |
*** rfolco has joined #openstack-dib | 12:07 | |
*** farosas has joined #openstack-dib | 12:07 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/diskimage-builder master: Updated from global requirements https://review.openstack.org/470451 | 12:13 |
*** isaacb has joined #openstack-dib | 13:29 | |
*** isaacb has quit IRC | 15:01 | |
*** hashar has quit IRC | 15:13 | |
*** rfolco has quit IRC | 15:56 | |
*** hashar has joined #openstack-dib | 15:57 | |
*** rfolco has joined #openstack-dib | 16:01 | |
*** hashar has quit IRC | 16:16 | |
*** jamielennox is now known as jamielennox|away | 16:26 | |
*** jamielennox|away is now known as jamielennox | 17:05 | |
*** chhavi has quit IRC | 17:55 | |
*** mordred has quit IRC | 18:47 | |
*** mordred has joined #openstack-dib | 18:48 | |
rfolco | ianw, around ? I think we should break down into 2 separate changes - https://review.openstack.org/447739 - 1) /dev/mapper change 2) ppc prep fix. Agree ? | 19:12 |
ianw | rfolco: hi | 19:12 |
ianw | rfolco: yes, that's pretty much where i'm heading. that review is out of date | 19:13 |
ianw | i'm trying to get consistent use of the dm nodes with https://review.openstack.org/#/c/471183/ | 19:13 |
ianw | currently we ... well, i'm not really sure ... we do "partprobe" on the loop device, and then check if the partitions exist, and if they don't we run kpartx, but don't check dm nodes ... it's all just very confused | 19:14 |
rfolco | ianw, if you don't mind, to keep changes simple, I would submit a new change for ppc fix because it does not have anything about /dev/mapper, or cleanups IMHO. | 19:20 |
ianw | rfolco: sure, you mean you have a fix? | 19:21 |
ianw | that doesn't rely on any of this? | 19:21 |
ianw | what i am thinking is that if we start passing the bootloader element the dm nodes, rather than /dev/loop directly, it "just works" | 19:22 |
rfolco | what would be the dm path ? /dev/mapper ? I tested it manually, it stills fails to run grub-install on it. | 19:23 |
rfolco | I have built manually grub with several commits, and after 2.02~beta3, it just works as is (but it is available on zesty only). So beta2 fails here: http://git.savannah.gnu.org/cgit/grub.git/tree/util/grub-install.c#n1771 | 19:25 |
rfolco | it apparently fails to open device and populate dev->disk->partition. | 19:25 |
ianw | rfolco: so i had this working | 19:25 |
ianw | yeah, that was some of the code i was looking at. if you go digging back, i'm not sure if i put comments in, or maybe if we check irc logs, i thought we found it taking different paths depending on the "/dev" arguments | 19:27 |
rfolco | prep is a just a dump of ELF into the partition, so all we need is to grub-mkimage and then dd the elf binary into the partition. | 19:27 |
rfolco | this is how I did for CirrOS image building | 19:28 |
rfolco | Bugs against xenial won't be accepted at this point, so the fix with mkimage and dd would be a good solution IMHO | 19:29 |
ianw | ok! if that works ... great :) | 19:31 |
ianw | so no grub-install step? | 19:31 |
ianw | i still think we want to use dm nodes anyway, because of LVM in the future | 19:32 |
rfolco | ianw, if ppc, mkimage and dd (no grub-install), yes. That's the only way i could fix it. I also made it work with zesty 2.02~beta3 grub2-common, but its ugly installing .deb directly, although it works. | 19:33 |
rfolco | ianw, question, when you say dm nodes, what would be the device path /dev/XXX? | 19:33 |
rfolco | /dev/mapper/? | 19:34 |
ianw | /dev/mapper/loop0p1 etc | 19:34 |
rfolco | I will confirm that, but still fails :) | 19:34 |
rfolco | ok, will make some more tests. Thanks for the braistorming, ianw | 19:35 |
ianw | rfolco: so per https://review.openstack.org/#/c/447739/22/diskimage_builder/elements/bootloader/finalise.d/50-bootloader we *had* it working | 19:36 |
ianw | using grub-install to "p2" (so not the "boot" partition, as you say maybe misnamed) | 19:36 |
ianw | and device mapper nodes | 19:36 |
rfolco | hmm, grub-install should target the prep partition, and only loads the remaning modules from $root/boot/grub (p2). Not sure this really worked. | 19:39 |
ianw | i'm sure mmedvede booted it | 19:40 |
ianw | i'm pretty sure, i mean :) | 19:40 |
rfolco | ok, will re-test it to see what happens | 19:41 |
ianw | rfolco: if you just take 447739 i guess it *should* work | 19:42 |
ianw | the problem there was, i just changed everything to use /dev/mapper/.... but didn't fix any of the cleanup | 19:43 |
mmedvede | rfolco: yeah, we got it working at some point with the grub that it installs | 19:43 |
mmedvede | I am 100% sure. It was not a pretty workaround to get it working, but it worked | 19:43 |
rfolco | mmedvede, did you boot it and grub loaded fine ? | 19:44 |
mmedvede | irrc, it booted into OS | 19:45 |
rfolco | ok | 19:45 |
rfolco | I will re-test it and let you guys know, thx | 19:46 |
mmedvede | rfolco: here is the log from when I tested the built image http://eavesdrop.openstack.org/irclogs/%23openstack-dib/%23openstack-dib.2017-03-22.log.html | 19:49 |
*** jamielennox has quit IRC | 20:34 | |
*** jamielennox has joined #openstack-dib | 20:38 | |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Start at block-device mounting/unmounting cleanup https://review.openstack.org/471183 | 20:50 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Start at block-device mounting/unmounting cleanup https://review.openstack.org/471183 | 20:54 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Start at block-device mounting/unmounting cleanup https://review.openstack.org/471183 | 21:13 |
*** farosas has quit IRC | 21:23 | |
*** pmannidi has joined #openstack-dib | 23:04 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!