| ianw | mmedvede: ok, thanks. | 00:01 |
|---|---|---|
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 00:21 |
| ianw | mmedvede: hmm, this looks wrong -> http://dal05.objectstorage.softlayer.net/v1/AUTH_3d8e6ecb-f597-448c-8ec2-164e9f710dd6/pkvmci/diskimage-builder/39/447739/1/check/dib-dsvm-functests-devstack-xenial/1889351/ | 00:28 |
| ianw | debian-minimal/stable-vm : not a known test (see -l) | 00:28 |
| mmedvede | ianw: yeah, I misunderstood how testcases are derived | 00:28 |
| mmedvede | ianw: I am fixing it now | 00:29 |
| ianw | cool, np :) | 00:29 |
| mmedvede | I have basically hardcoded the list of testcases | 00:29 |
| *** pmannidi has quit IRC | 00:51 | |
| *** pmannidi has joined #openstack-dib | 00:52 | |
| *** jamielennox is now known as jamielennox|away | 01:16 | |
| *** pleia2_ is now known as pleia2 | 01:25 | |
| *** jamielennox|away is now known as jamielennox | 01:28 | |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 01:31 |
| mmedvede | ianw: the ppc test succeeded on ^, but it did not actually test the failure mode on Xenial. I am switching back to debian, disabling xenial. Can not have both enabled because ubuntu uses non-default distro mirror url on ppc | 01:41 |
| ianw | hmm, ok | 03:11 |
| *** jamielennox is now known as jamielennox|away | 04:12 | |
| *** pmannidi has quit IRC | 05:00 | |
| *** pmannidi has joined #openstack-dib | 05:01 | |
| *** aparnav has joined #openstack-dib | 05:27 | |
| *** yolanda has quit IRC | 06:32 | |
| *** yolanda has joined #openstack-dib | 06:46 | |
| *** andreas-f has quit IRC | 06:48 | |
| *** hashar has joined #openstack-dib | 08:45 | |
| *** openstackgerrit has quit IRC | 09:03 | |
| *** pmannidi has quit IRC | 09:09 | |
| *** pmannidi has joined #openstack-dib | 09:10 | |
| *** jamielennox|away is now known as jamielennox | 10:30 | |
| *** jamielennox is now known as jamielennox|away | 11:05 | |
| *** jamielennox|away is now known as jamielennox | 11:12 | |
| yolanda | ianw, greghaynes , any idea why https://review.openstack.org/#/c/447047/6/diskimage_builder/elements/dracut-regenerate/finalise.d/98-dracut-regenerate is not installing the packages i need? | 11:25 |
| yolanda | i check logs and i see it enters on the install-packages part | 11:26 |
| yolanda | but then i inspect the image,and i don't have lvm2 or cryptsetup installed | 11:26 |
| yolanda | i added some logging, and there are no errors on the install part | 11:26 |
| *** yolanda has quit IRC | 11:52 | |
| *** yolanda has joined #openstack-dib | 11:54 | |
| *** openstackgerrit has joined #openstack-dib | 12:01 | |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Add dracut-regenerate elements https://review.openstack.org/447047 | 12:01 |
| *** radez` is now known as radez | 12:01 | |
| *** aparnav has quit IRC | 12:34 | |
| *** chhavi has joined #openstack-dib | 13:16 | |
| *** pmannidi has quit IRC | 13:18 | |
| *** pmannidi has joined #openstack-dib | 13:19 | |
| openstackgerrit | Mikhail S Medvedev proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 13:39 |
| *** dmarlin has joined #openstack-dib | 14:18 | |
| *** hashar has quit IRC | 14:33 | |
| *** hashar has joined #openstack-dib | 14:35 | |
| *** venki has joined #openstack-dib | 15:16 | |
| venki | Hi guys... | 15:16 |
| venki | Can i create deploy-kernal and deploy-initrd (ironic configuration) for Android OS ? | 15:17 |
| *** chhavi has quit IRC | 15:28 | |
| *** chhavi has joined #openstack-dib | 15:49 | |
| *** makowals_ has joined #openstack-dib | 16:15 | |
| *** chhavi has quit IRC | 17:10 | |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Add dracut-regenerate elements https://review.openstack.org/447047 | 17:20 |
| *** hashar has quit IRC | 17:28 | |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Add dracut-regenerate elements https://review.openstack.org/447047 | 18:55 |
| ianw | mmedvede: oh good, that vm element test at least seems to get us further | 19:05 |
| ianw | venki: android ... i don't believe there's been any work on that. anything's possible i guess | 19:06 |
| ianw | yolanda: hmm, that would not exit with any failure code if the subprocess was creating one? hiding the failure? | 19:08 |
| yolanda | ianw, well, the package install stuff is solved now, i was using install-packages -m <package_name>, i removed the -m flag and worked | 19:13 |
| yolanda | however, i'm debugging a strange failure now. When regenerating dracut, something happens that causes conflict on iscsi | 19:14 |
| yolanda | the vm cannot boot with an error about "open-iscsi service failed" | 19:14 |
| ianw | ohh, that "-m" issue, i have deja vu about that | 19:14 |
| yolanda | what shall -m do? i copied from some other element, but i realized i was wrong | 19:15 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 19:20 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 19:31 |
| ianw | yolanda: runs the mapping bit iirc | 19:36 |
| ianw | yolanda: sorry, not much help with iscsi type things :( | 19:36 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 19:39 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 19:39 |
| *** hashar has joined #openstack-dib | 19:53 | |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 19:59 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 20:00 |
| mmedvede | ianw: I think I know what the regression is. https://review.openstack.org/#/c/425824/6/diskimage_builder/lib/disk-image-create : if IMAGE_BLOCK_DEVICE does not exist, it still ran 'kpartx' if it was on ppc. While in 2.1.0 you get this: https://github.com/openstack/diskimage-builder/blob/52b16b9883ccbb7febaa8fa785cc56e036d86668/diskimage_builder/block_device/level1/partitioning.py#L188 | 20:06 |
| mmedvede | so with the new code, 'kpartx -avs' would never run on ppc if the partition exists | 20:06 |
| ianw | mmedvede: as a first thing, i believe we've dropped actually creating the prep boot partition originally in https://git.openstack.org/cgit/openstack/diskimage-builder/tree/elements/vm/block-device.d/10-partition?h=1.28.0#n13 | 20:07 |
| mmedvede | ianw: ok, I missed that | 20:09 |
| ianw | mmedvede: i think now we're creating the partitions -> http://dal05.objectstorage.softlayer.net/v1/AUTH_3d8e6ecb-f597-448c-8ec2-164e9f710dd6/pkvmci/diskimage-builder/99/448099/4/check/dib-dsvm-functests-devstack-xenial/d9bc6a2/ | 20:13 |
| mmedvede | yup, was just looking there \o/ | 20:13 |
| mmedvede | back to the original error | 20:13 |
| ianw | that's kind of good ... i think we just need to make sure grub is running against the first partition | 20:15 |
| mmedvede | ianw: I guess adding kpartx back in might fix it? | 20:15 |
| mmedvede | if I talk nonsense, let me know :) | 20:17 |
| ianw | mmedvede: hmm, i think what happens is we set IMAGE_BLOCK_DEVICE to the root partition to be compatible with old versions | 20:21 |
| ianw | but PPC was basically overriding this | 20:21 |
| mmedvede | ianw: ack | 20:22 |
| ianw | https://git.openstack.org/cgit/openstack/diskimage-builder/tree/elements/vm/block-device.d/10-partition?h=1.28.0#n51 | 20:22 |
| * mmedvede still over his head on all the partitioning business | 20:22 | |
| ianw | where it sets IMAGE_BLOCK_DEVICE to p2 | 20:22 |
| ianw | https://git.openstack.org/cgit/openstack/diskimage-builder/tree/elements/bootloader/finalise.d/50-bootloader?h=1.28.0#n15 | 20:26 |
| ianw | yeah, then it overrides it again | 20:26 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 20:30 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 20:31 |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: WIP: Add lvm management to diskimage-builder https://review.openstack.org/444403 | 20:34 |
| openstackgerrit | Mikhail S Medvedev proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 20:43 |
| openstackgerrit | Mikhail S Medvedev proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 20:44 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 21:05 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 21:06 |
| ianw | ^ this one hopefully gives us some more info ... | 21:06 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 21:13 |
| ianw | 2017-03-21 21:11:02.585 | [debian-minimal/stable-vm] Device Boot Start End Sectors Size Id Type | 21:13 |
| ianw | 2017-03-21 21:11:02.585 | [debian-minimal/stable-vm] /dev/loop0p1 * 2048 206847 204800 100M 41 PPC PReP Boot | 21:13 |
| ianw | 2017-03-21 21:11:02.586 | [debian-minimal/stable-vm] /dev/loop0p2 206848 1191551 984704 480.8M 83 Linux | 21:13 |
| ianw | i think getting closer ... | 21:13 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 21:14 |
| *** andreas-f has joined #openstack-dib | 21:22 | |
| ianw | ahh, so prep is actually a flag, like bootable flag | 21:30 |
| * mmedvede reads | 21:42 | |
| *** dmarlin has quit IRC | 22:22 | |
| ianw | -000001c0 02 00 41 03 20 1f 01 00 00 00 ff 0f 00 00 00 00 |..A. ...........| | 22:40 |
| ianw | +000001c0 02 00 83 03 20 1f 01 00 00 00 ff 0f 00 00 00 00 |.... ...........| | 22:40 |
| ianw | not quite true ... | 22:40 |
| ianw | "set 1 prep on" just turns the filesystem type to prep (0x41 in the 3rd byte there) | 22:41 |
| ianw | that's the only difference | 22:41 |
| ianw | ergo, we seem to be doing that, but it still doesn't like the partition | 22:41 |
| bkero | What partition table type are you using? | 22:43 |
| ianw | bkero: see 2017-03-21 21:19:22.010 in http://dal05.objectstorage.softlayer.net/v1/AUTH_3d8e6ecb-f597-448c-8ec2-164e9f710dd6/pkvmci/diskimage-builder/99/448099/8/check/dib-dsvm-functests-devstack-xenial/88860b7/ | 22:47 |
| ianw | we are correctly setting it to prep | 22:47 |
| bkero | ianw: have you run yaboot and diffed the disk to see what that does? | 22:49 |
| ianw | bkero: we're before that ... | 22:51 |
| ianw | but hang on, i think grub-install wants to work on the whole disk, and finds a prep partition ... | 22:51 |
| bkero | /usr/sbin/grub-install: error: the chosen partition is not a PReP partition. | 22:52 |
| bkero | yeah | 22:53 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 22:53 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 22:53 |
| bkero | Here's how grub determines if the partition is a prep partition: https://github.com/coreos/grub/blob/master/util/grub-install.c#L697 | 22:56 |
| bkero | I'm assuming that '00 41' in your hexdump is 0x41. I'm assuming Disklabel type: dos is synonymous with 'msdos' | 22:57 |
| ianw | bkero: yeah, i don't think it's an issue with disk labels. but i think what's happening is it sees *if* there is a prep partition, and that will be "install_drive" in that file | 22:58 |
| bkero | ianw: can follow the logic here: https://github.com/coreos/grub/blob/master/util/grub-install.c#L1760 | 22:59 |
| ianw | the thing is where it sets install_drive | 23:00 |
| ianw | install_device i mean | 23:00 |
| ianw | well it doesn't like just /dev/loop0 either | 23:04 |
| bkero | of course not | 23:08 |
| bkero | ianw: I wonder if !dev->disk->partition is returning 0 because it's failing to identify partition type of p2 | 23:09 |
| bkero | hahahaha wow look at this | 23:11 |
| bkero | https://github.com/coreos/grub/blob/grub-2.02-beta2/grub-core/osdep/linux/ofpath.c#L517 | 23:12 |
| bkero | I didn't realize the 'device' string was 'sda1'. | 23:13 |
| bkero | Looks like it's returning NULL because the device name is 'loop0p1' | 23:14 |
| ianw | yeah i saw that, but i think before we were using /dev/mapper which it also doesn't seem to handle | 23:14 |
| bkero | grub-probe is likely being called before grub-install and handing it some bogus data | 23:15 |
| ianw | mmedvede: how did we go with an old build log? that would be quite handy at this point | 23:16 |
| mmedvede | ianw: the last working build was using IMAGE_BLOCK_DEVICE=/dev/mapper/loop0p2 | 23:17 |
| ianw | 2017-03-21 22:58:14.744 | [debian-minimal/stable-vm] Installation finished. No error reported. | 23:17 |
| mmedvede | which was mapped with kpartx | 23:17 |
| ianw | see that in | 23:17 |
| ianw | http://dal05.objectstorage.softlayer.net/v1/AUTH_3d8e6ecb-f597-448c-8ec2-164e9f710dd6/pkvmci/diskimage-builder/99/448099/9/check/dib-dsvm-functests-devstack-xenial/4e4cfcd/ | 23:18 |
| ianw | that's interesting, that just "grub-install" called from the post of the .deb install i guess | 23:18 |
| mmedvede | ianw: the old 10-partition log http://paste.openstack.org/show/603705/ | 23:23 |
| ianw | hmmm, what really is the difference between the loopback device and the mapper device? | 23:24 |
| ianw | mmedvede: can you find in the logs what the argument to grub-install is then? | 23:25 |
| ianw | is it /dev/mapper/loop0p2? | 23:25 |
| mmedvede | + /usr/sbin/grub-install --modules=part_msdos --force /dev/mapper/loop0p1 --no-nvram | 23:25 |
| ianw | ok ... and that's the prep partition | 23:27 |
| ianw | that's what i was trying and still getting the failure | 23:27 |
| mmedvede | but from what we see, it does not seem like grub-install respects the arg anyway | 23:28 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 23:31 |
| ianw | ^ run kpartx and see if install to /dev/mapper works? | 23:31 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 23:31 |
| ianw | mmedvede: can you pastebin the bits around that grub-install. do we get the "unknown device" message still? | 23:32 |
| mmedvede | worth a try. Also might have to set one of device path vars to /dev/mapper/loop0p2. Not sure if kpartx on its own would change anything | 23:34 |
| *** jamielennox is now known as jamielennox|away | 23:35 | |
| mmedvede | ianw: there is no warning, http://paste.openstack.org/show/603706/ | 23:35 |
| ianw | the original change doesn't explain why it uses /dev/mapper | 23:35 |
| ianw | https://review.openstack.org/#/c/153404/ | 23:35 |
| mmedvede | ianw: maybe just copy-paste from the non-ppc branch | 23:36 |
| mmedvede | why kpartx was used in the first place? | 23:36 |
| ianw | ? is Yogananth still around? | 23:37 |
| mmedvede | let me check | 23:38 |
| *** jamielennox|away is now known as jamielennox | 23:39 | |
| ianw | arggh, the kpartx is going to stdout and messing up | 23:44 |
| *** cloudkiller has joined #openstack-dib | 23:49 | |
| ianw | it does seem that grub probing has a few different paths for /dev/mapper | 23:49 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Create PReP boot partition for PPC https://review.openstack.org/447739 | 23:49 |
| ianw | one of them might be important | 23:50 |
| mmedvede | if he is still at IBM, Yogananthan is not active | 23:50 |
| *** cloudnull has quit IRC | 23:50 | |
| *** cloudkiller is now known as cloudnull | 23:50 | |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: [DNM] Test boot partition code https://review.openstack.org/448099 | 23:50 |
| mmedvede | original reason for using kpartx: "If the loop partition isn't created as a device by the kernel, go ahead and use kpartx to create the device.". After that it got copied as more special cases where added. Maybe ppc happened to work by luck due to how grub probe works | 23:54 |
| ianw | it does seem there was an unfortunate amount of luck holding this all together :) | 23:55 |
| mmedvede | evolution in action | 23:56 |
| *** hashar has quit IRC | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!