[00:02] <CIA-9> ubiquity: cjwatson * r3271 ubiquity/debian/changelog: releasing version 1.13.1
=== mcasadevall is now known as NCommander
[08:21] <evand> cjwatson_: Is it not worth the overhead to check the value of partman/default_filesystem for those cases in ubiquity?
[09:51] <cjwatson_> evand: there's only one where it really matters (the others were just cosmetic) - we probably should check default_filesystem for it once at startup, though, feel free :)
[09:52] <evand> will do
=== cjwatson_ is now known as cjwatson
[10:03] <corecode> hey
[10:03] <corecode> is there a way to write a partman receipe that spans multiple disks?
[10:04] <corecode> ah, device
[10:04] <corecode> i see
[10:06] <corecode> then i wonder, how can i list all available drives so that i can put them into partman-auto/disk
[10:09] <evand> cjwatson: when you say "at startup", do you mean ubiquity startup or partman startup?
[10:21] <cjwatson> I don't suppose it matters too much
[10:21] <cjwatson> I was thinking the former
[10:22] <cjwatson> corecode: if you're on at least intrepid, you can use a combination of debconf-set and list-devices in partman/early_command to set that dynamically
[10:22] <corecode> list-devices, is that only available in the installer?
[10:22] <cjwatson> (before intrepid, partman/early_command wasn't available, so there was no easily available scripting hook that ran after disks were detected)
[10:22] <cjwatson> yes
[10:22] <corecode> okay
[10:23] <cjwatson> 'list-devices disk'
[10:23] <cjwatson> prints '/dev/sda' here
[10:23] <corecode> ah great
[10:23] <corecode> thanks
[10:24] <corecode> now i just need to write a nice receipe to create a pv on the second drive and add it to the vg
[10:24] <corecode> recipe*
[11:25] <corecode> hm
[11:25] <corecode> parted_server: OUT: Can't have a partition outside the disk!
[11:25] <corecode> parted_server: OUT: Can't have a partition outside the disk!
[11:25] <corecode> sorry
[11:25] <corecode> how am i supposed to tell partman "create a primary spanning the whole disk as pv?
[11:28] <corecode> i wrote in the recipe:  100 1000 -1 $default_filesystem $defaultignore{ } $primary{ } method{ lvm } device{ /dev/sdb }
[11:28] <corecode> but i guess the -1 does some wrong calculations
[11:29] <corecode> because parted does this:
[11:29] <corecode> parted_server: add_primary_partition(disk(312500000),0-312500000)
[11:29] <corecode> /bin/autopartition-lvm: IN: NEW_PARTITION =dev=sdb primary ext3 0-159999999999 beginning 160000000001
[11:33] <cjwatson> I'm afraid I'm doing several other things right now, but maybe you could file a bug on partman-auto-lvm attaching syslog and partman logs
[11:34] <cjwatson> your recipe *ought* to work, I agree
[11:34] <corecode> mhm
[11:49] <corecode> https://bugs.launchpad.net/ubuntu/+source/partman-auto/+bug/368924
[11:49] <ubottu> Launchpad bug 368924 in partman-auto ""Can't have a partition outside the disk!" error when two partitions are flagged as $primary in a partman-auto/expert_recipe" [Undecided,New]
[11:49] <corecode> somebody already filed that report
[11:53] <cjwatson> err
[11:53] <cjwatson> you sure yours is the same? doesn't sound like it to me
[11:53] <cjwatson> same error message doesn't imply same bug
[11:54] <corecode> yah
[11:54] <corecode> testing
[11:54] <cjwatson> in any case I'd appreciate a separate bug
[11:54] <cjwatson> it's much easier to mark a bug as a duplicate than it is to split some of the comments out of a single bug
[12:36] <CIA-9> partman-basicfilesystems: cjwatson * r563 hardy-proposed/debian/changelog: releasing version 56ubuntu5
[14:39] <cody-somerville> Would it be possible for partman to partition the disk based off of the sizes of certain files on the install media?
[14:40] <cody-somerville> or my real question is, is it possible to do it relatively easily?
[14:40] <cjwatson> you could write a preseed file to fetch that information and write out a custom recipe
[14:40] <cjwatson> I'd rather not hack that into partman itself
[14:41] <cody-somerville> cjwatson, So use early_command to execute a script to calculate the size of the utility/recovery partition and write up a recipe at that time?
[14:41] <cjwatson> right
[14:47] <corecode> cjwatson: will parted-auto-lvm remove existing partitions on a non-primary device?
[14:48] <cjwatson> "non-primary device"?
[14:48] <corecode> not the first disk device
[14:50] <cjwatson> no, it'll only operate on the device you tell it to operate it
[14:50] <corecode> but auto_lvm_prepare can work on more than one device
[14:51] <corecode> what i want to say is, where do all existing partitions get deleted before applying a recipe?
[14:52] <cjwatson> auto_lvm_prepare only works on more than one device if you tell it to, which we don't
[14:52] <cjwatson> anyway, if you're already reading the code ... :)
[14:52] <corecode> yah, i'm trying to find the problem
[14:52] <corecode> i think it is because there is already a partition on the destination (secondary) drive
[14:56] <cody-somerville> cjwatson, I'm thinking maybe a udeb that runs before partman would be better/more flexible than exploiting early_command
[14:57] <cjwatson> corecode: auto_lvm_prepare calls auto_init_disks (defined in partman-auto)
[14:57] <cjwatson> cody-somerville: up to you, I'd do it with an early_command myself and I think that's pretty much equally flexible
[14:57] <cjwatson> partman/early_command that is
[14:57] <cjwatson> well, on >= intrepid
[14:58] <cody-somerville> cjwatson, I already use early_command and the nice thing about the udeb is that it just needs to get included and it'll work instead of me modifying the preseed file for all our jaunty projects
[14:59] <cjwatson> ok
[15:00] <corecode> cjwatson: then this can't be the problem...
[15:00] <cjwatson> corecode: your use of "destination" and "secondary" is confusing me
[15:01] <cjwatson> perhaps it would help to use concrete examples
[15:01] <corecode> i have sda and sdb
[15:01] <corecode> i want to create two pv and one vg
[15:01] <corecode> now just sda works fine, but when i try to get partman to create an pv on sdb, it fails
[15:02] <cjwatson> ah, now I discover I was looking at old partman-auto code ...
[15:03] <cjwatson> so my previous statement that we never tell auto_lvm_prepare to work on more than one device was wrong, sorry about that
[15:03] <cjwatson> it should operate basically the same way on all the disks you give it, from what I can see
[15:07] <corecode> i think the problem is that it can't allocate a primary partition that spans all drive
[15:10] <cjwatson> what's the exact error message?
[15:10] <corecode> /bin/autopartition-lvm: IN: NEW_PARTITION =dev=sdb primary ext3 0-159999999999 beginning 160000000001
[15:10] <corecode> parted_server: add_primary_partition(disk(312500000),0-312500000)
[15:10] <corecode> parted_server: OUT: Can't have a partition outside the disk!
[15:11] <cjwatson> oh, you said, sorry
[15:11] <cjwatson> it sounded to me as if your initial guess that it was an arithmetic error somewhere is right
[15:12] <cjwatson> did you file a bug with the logs?
[15:12] <corecode> no, i wanted to see the bug first
=== robbiew1 is now known as robbiew
[15:52] <corecode> cjwatson: i think it is an alignment problem
[15:52] <corecode> cjwatson: within parted
[15:53] <corecode> cjwatson: or however parted gets driven
[15:53] <cjwatson> corecode: I really can't comment any further until I see logs, sorry
[15:53] <cjwatson> I'd just be guessing
[15:54] <corecode> really, the logs wouldn't help
[15:55] <cjwatson> then I can't help you at all, since often the logs help *me*
[15:55] <cjwatson> I have a lot of practice at using partman logs to trace through what it's doing
[15:56] <corecode> what's the scp anna package?
[15:56] <cjwatson> openssh-client-udeb
[15:56] <corecode> thx
[16:01] <corecode> http://chlamydia.fs.ei.tum.de/~corecode/unsorted/partman
[16:02] <cjwatson> could you file a bug and attach it?
[16:02] <cjwatson> I'll lose it otherwise :)
[16:02] <corecode> i don't know what to write in the bug
[16:03] <corecode> doesn't work seems a bit vague
[16:03] <cjwatson> "partman-auto-lvm fails to create PV on second disk: 'Can't have a partition outside the disk!'"
[16:03] <cjwatson> or something like that
[16:03]  * cjwatson -> meeting
[16:04] <cjwatson> my guess is it's a fencepost error - 160000000000 is a suspiciously round number for the size of the disk
[16:06] <corecode> must be an error in parted_server then
[16:06] <cjwatson> not necessarily, it's fed the sizes externally
[16:06] <cjwatson> all the shell bits of partman do a lot of size computations
[16:38] <corecode> hm, could have been pilot error.  i forgot a .
[16:38] <corecode> but now it tells me "no root file system defined"
[16:59] <CIA-9> debian-installer: cjwatson * r1104 ubuntu/ (5 files in 2 dirs): Move mainline architectures to 2.6.30-8 kernels.
[17:03] <CIA-9> debian-installer: cjwatson * r1105 ubuntu/debian/changelog: releasing version 20081029ubuntu40
[18:55] <corecode> cjwatson: yep, whenever the system tries to create a big primary partition, it breaks
[20:43] <cody-somerville> If I use oem-config-udeb, do I have to preseed a late command to run oem-configure-prepare in the target to skip the "boot and customize and then click the desktop icon to be ready for the 'first' boot"?
[20:56] <CIA-9> usb-creator: rgreening * r106 usb-creator/ (4 files in 3 dirs): (log message trimmed)
[20:56] <CIA-9> usb-creator: - Sort of have kdesudo working. Still doesn't pick up cmdline args if re-launched.
[20:56] <CIA-9> usb-creator: - Replaced kde_frontend with
[20:56] <CIA-9> usb-creator:  bzr+ssh://bazaar.launchpad.net/~david.edmundson/usb-creator/kde-frontend/
[20:56] <CIA-9> usb-creator:  Using this as a closer starting point for the KDE/Qt conversion...
[20:56] <CIA-9> usb-creator: - Backed up old kde_frontend.py to kde_frontend.py.old
[20:56] <CIA-9> usb-creator: - Started updating and implementing missing bits in kde_frontend.py
[22:13] <cjwatson> cody-somerville: you'd have to, yes
[22:13] <cjwatson> (though I don't recall testing that method at all recently ...)