*** phuongnh has joined #openstack-dib | 01:14 | |
*** jesusaur has quit IRC | 01:45 | |
*** jesusaur has joined #openstack-dib | 01:53 | |
openstackgerrit | Tristan Cacqueray proposed openstack/diskimage-builder master: rpm-distro: set the contentdir yum var https://review.openstack.org/568180 | 03:31 |
---|---|---|
*** greghaynes has quit IRC | 05:16 | |
*** greghaynes has joined #openstack-dib | 05:24 | |
openstackgerrit | Tristan Cacqueray proposed openstack/diskimage-builder master: rpm-distro: set the contentdir yum var https://review.openstack.org/568180 | 05:39 |
*** eshas has joined #openstack-dib | 06:02 | |
*** eshas has quit IRC | 06:03 | |
*** andreas_s has joined #openstack-dib | 06:28 | |
*** pavel_V has joined #openstack-dib | 07:32 | |
openstackgerrit | Merged openstack/diskimage-builder master: Trivial: update url to new url https://review.openstack.org/568117 | 08:06 |
openstackgerrit | Merged openstack/diskimage-builder master: Fixes add-apt-keys in dpkg element https://review.openstack.org/565599 | 08:52 |
openstackgerrit | Merged openstack/diskimage-builder master: Add pip cache cleanup to pip-and-virtualenv https://review.openstack.org/562055 | 09:03 |
olivierb- | Hello dibers, what is the difference/relashioship/... between the DIB_BLOCK_DEVICE environment variable contents and the label entry of the partitioning section when using custom partitioning ? | 09:58 |
*** sshnaidm|rover has joined #openstack-dib | 10:49 | |
sshnaidm|rover | DIB cores, please review/merge: https://review.openstack.org/#/c/568180/ it's urgently needed for tripleo | 10:49 |
ianw | olivierb-: what do you mean label entry? | 10:59 |
olivierb- | ianw https://github.com/openstack/diskimage-builder/blame/master/doc/source/user_guide/building_an_image.rst#L318 | 11:02 |
ianw | olivierb-: oh right ... yeah i will admit things are a bit confused. i'm not sure i *really* meant DIB_BLOCK_DEVICE to be user set | 11:04 |
ianw | the idea was to include one of the block-device-[mbr|efi|gpt] elements, which sets it for you *and* provides a yaml config file with the label, etc | 11:04 |
olivierb- | however for vm element /block-device-* seems to set this variable too | 11:05 |
olivierb- | adding to the confusion | 11:05 |
ianw | and my thinking was that you could write "block-device-olivierb" (or whatever) that implements whatever way you want to configure the disk | 11:05 |
olivierb- | currently I do not use vm element as I need baremetal image and was using DIB_BLOCK_DEVICE=mbr with label: mbr in my custom (LVM based) partitioning which worked great | 11:06 |
olivierb- | I am now trying to migrate to gpt but no luck so far, can not even generate the image | 11:07 |
ianw | olivierb-: hmm, yes in that case, your custom partitioning would require a "gpt" label | 11:08 |
ianw | it's two sort of separate things. the yaml configuration file is read by dib-block-device tool, which creates the actual image and sets the partition types, etc | 11:09 |
ianw | DIB_BLOCK_DEVICE=gpt then instructs diskimage_builder/elements/bootloader/finalise.d/50-bootloader to install grub for gpt | 11:09 |
*** andreas_s has quit IRC | 11:10 | |
*** andreas_s has joined #openstack-dib | 11:10 | |
ianw | I actually agree that i think this is a bit wrong. I *think* that 50-bootloader could actually look at the label that is setup in the configuration file | 11:10 |
ianw | i would be very willing to review patches to that effect :) | 11:10 |
ianw | it would need to be set in a similar way to IMAGE_BLOCK_DEVICE | 11:12 |
*** andreas_s has quit IRC | 11:15 | |
olivierb- | got it, many thanks, will try to submit patches soon regarding this | 11:16 |
olivierb- | may be yolanda_ who helped me getting my first mbr based, lvm customized image to work already tried GPT too ??? | 11:18 |
olivierb- | df | 11:18 |
olivierb- | sorry wrong window ;-) | 11:18 |
*** phuongnh has quit IRC | 11:18 | |
olivierb- | for df I meant | 11:19 |
ianw | olivierb-: yeah, looking behind how IMAGE_BLOCK_DEVICE works with the call to "dib-block-device getval ..." and then exporting a flag is what i'm thinking | 11:20 |
ianw | however, we still need to have default custom configuration files; like if someone wants efi they shouldn't have to write their own yaml with the efi partition etc | 11:21 |
sshnaidm|rover | yolanda_, can you please take a look? https://review.openstack.org/#/c/568180/ | 11:23 |
*** needsleep is now known as TheJulia | 11:25 | |
*** andreas_s has joined #openstack-dib | 11:26 | |
yolanda_ | hi sshnaidm|rover taking a look | 11:40 |
olivierb- | hello yolanda_ , did you see the exchanges above concerning mbr vs gpt ??? | 11:41 |
yolanda_ | olivierb-, just read those. As ianw said, you need to specify now DIB_BLOCK_DEVICE=xxx, to discriminate between mbr/gpt | 11:42 |
olivierb- | ianw yolanda_ just made some more trials on my side and found something weird which requires your expert eyes. Without changing my custom partitioning yaml, I just changed DIB_BLOCK_DEVICE=mbr to DIB_BLOCK_DEVICE=gpt in my shell environment an regenerated my image. Image is same than with mbr and trace shows that only what is in DIB_BLOCK_DEVICE_CONFIG seems to be parsed (INFO diskimage_builder.block_device.level1.mbr [-] Create MBR | 11:48 |
olivierb- | disk partitioning object) | 11:48 |
olivierb- | so seems like DIB_BLOCK_DEVICE_CONFIG yaml contents has precedence over DIB_BLOCK_DEVICE | 11:49 |
olivierb- | is this expected ? | 11:49 |
yolanda_ | olivierb-, so the bootloader element just takes DIB_BLOCK_DEVICE var, anything else | 11:51 |
yolanda_ | but you need to change both vars to be the same | 11:52 |
yolanda_ | i mean, both DIB_BLOCK_DEVICE and DIB_BLOCK_DEVICE_CONFIG should have the same partition time | 11:52 |
yolanda_ | type | 11:52 |
yolanda_ | an improvement should be that we don't need that DIB_BLOCK_DEVICE var anymore, and that we read the type from the DIB_BLOCK_DEVICE_CONFIG yaml file, in the bootloader element | 11:52 |
olivierb- | yes, I agreed with ianw to work on this soon | 11:53 |
yolanda_ | thx! | 11:53 |
olivierb- | for the gpt case, currently trying to change both but got 1st issue because says that primary flag is not supported for GPT (my bad) so retrying now | 11:55 |
olivierb- | second error was because I did not create the BIOS partition => no embedding possible but the message is very clear and got the infos from the dib docs on how to solve this => great, retrial ongoing | 12:08 |
olivierb- | now working like a charm, thanks ianw and yolanda_ | 12:11 |
yolanda_ | nice | 12:23 |
openstackgerrit | Merged openstack/diskimage-builder master: rpm-distro: set the contentdir yum var https://review.openstack.org/568180 | 12:36 |
*** vmlinuz has joined #openstack-dib | 12:39 | |
*** mjturek has joined #openstack-dib | 13:03 | |
olivierb- | just another question on gpt generated dib image which is then provisioned to a machine using bifrost/ironic. When I launch fdisk on the disk where the image is written, I get: GPT PMBR size mismatch (125829119 != 209715199) will be corrected by w(rite).(message is repeated twice). However when I launch a KVM/libvirt instance using the qcow2 directly I do not get any error/warning. Does this ring a bell to someone ??? | 13:55 |
olivierb- | should I worry about this error, my gut feeling tells me I should | 13:55 |
*** rajinir has joined #openstack-dib | 14:40 | |
clarkb | olivierb-: not familiar to me, but not using the images on baremetal. Perhaps something to do with hardware block sizes? | 15:29 |
*** vmlinuz has quit IRC | 15:40 | |
*** andreas_s has quit IRC | 15:48 | |
*** vmlinuz has joined #openstack-dib | 15:52 | |
*** vmlinuz has joined #openstack-dib | 15:52 | |
*** andreas_s has joined #openstack-dib | 16:00 | |
*** andreas_s has quit IRC | 16:06 | |
*** sshnaidm|rover is now known as sshnaidm|off | 18:07 | |
*** Guest16323 is now known as mgagne | 18:40 | |
*** mgagne has joined #openstack-dib | 18:41 | |
*** vmlinuz has quit IRC | 19:00 | |
*** vmlinuz has joined #openstack-dib | 19:01 | |
*** Sukhdev has joined #openstack-dib | 20:55 | |
*** sshnaidm|off has quit IRC | 21:16 | |
*** mjturek has quit IRC | 21:40 | |
*** vmlinuz has quit IRC | 23:10 | |
*** rajinir has quit IRC | 23:19 | |
*** vmlinuz has joined #openstack-dib | 23:27 | |
*** vmlinuz has joined #openstack-dib | 23:27 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!