March 2010 - Arch-releng - lists.archlinux.org

Re: [arch-releng] [arch-general] [patch] AIF, discover repos on iso.
by Dieter Plaetinck 05 Mar '10

05 Mar '10

On Fri, 19 Feb 2010 22:26:54 +0100 (CET) Mark Pustjens <pustjens(a)dds.nl> wrote: > Hi List, > > The attached patch fixes a TODO in lib-pacman of aif. > Instead of hardcoding that `core' is always available locally and > falling back to net for others, it now checks /src/ for repos. > > It assumes repos are stored at /src/$repo/pkg/. All these repos are > added as cache dirs. > > While preparing pacman, a number of repo names are provided to be > prepared. > > All those repos which were requested are added as actual repo. If > this repo is available locally, the url points to disk. Otherwise it > falls back to the mirrorlist. > > Please let me know what you think. > > Greetings/Groetjes > > Mark Pustjens > thanks. it looks pretty good. i like how you automatically add the right cachedirs. However, when adding repositories to the pacman config, * i would not do "Server = file:///src/${repo}/pkg" just because /src/repo/pkg exists. what if the user has an outdated core image and wants to do a netinstall? he might have explicitly configured a mirror (SYNC_URL), but it would get ignored like this. * in prefill_configs(), we only add an entry to the mirrorlist when [ "$var_PKG_SOURCE_TYPE" = "net" -a -n "${var_SYNC_URL}" ] i.e. with the following config: SOURCE=cd FILE_URL=file:///src/core/pkg nothing is added to $var_MIRRORLIST which is probably a good thing because we don't want `file:///src/core/pkg`-eqsue urls on the target system. Since you point to $var_MIRRORLIST instead of $serverurl for repo's which are not available in /src/ at all, this means they must be configured as netinstall. So if the user wants a netinstall, it's his own reponsability to fill in SYNC_URL correctly. however, this also implies if he wants to mount a custom package repository, he _must_ do it under /src/<reponame>, not any other dir. otherwise the code won't see it and use $var_MIRRORLIST which may contain a mirror not containing <reponame>, or even nothing at all. basically, the way i see it is this: - setting cachedirs for repos in /src never hurts - SYNC_URL set and netinstall: use $var_MIRRORLIST for all requested repo's, even if they are in /src - netinstall but -z $SYNC_URL: same as above, it's the users responsability to set SYNC_URL - cd install and FILE_URL not set: check if the repo is in /src, if so add it. if not, don't add it to pacman.conf at all - cd install and FILE_URL set: same as above but try to use FILE_URL first (*). I've also been thinking if things should not be more configurable then this. (for example, user wants to use core in /src, but extra from SYNC_URL (**)), but per-repository settings seem more complicated so I hope they are not needed. Actually, since we only support "updating all packages simultaneously only" anyway, the per-repo configurability might be to be avoided by definition. I hope you understand what i mean. i put some thought into this and it's getting late, so .. let me know. and if you want to update your patch, it would be much more convenient if you could host a git clone somewhere i could pull from. Dieter (*): we should probably only let the user specify /src himself and add /$repo/pkg ourself in the automatic procedure. as the '/pkg' part probably never changes?? otherwise the user must specify something like '/src/$repo/pkg' and we replace $repo. (**): if you think wanting to use any other repo then core is an overcomplication and you can install extra packages after rebooting, don't forget about automatic installations where you want to install all needed packages at once.

1 0

Re: [arch-releng] [arch-general] [patch] AIF, partition table compatibility with grub2.
by Dieter Plaetinck 05 Mar '10

05 Mar '10

On Wed, 24 Feb 2010 16:02:53 -0600 Aaron Griffin <aaronmgriffin(a)gmail.com> wrote: > On Wed, Feb 24, 2010 at 2:37 PM, Dieter Plaetinck > <dieter(a)plaetinck.be> wrote: > > On Tue, 23 Feb 2010 00:18:28 +0100 > > Jan de Groot <jan(a)jgc.homeip.net> wrote: > > > >> On Mon, 2010-02-22 at 23:21 +0100, Tobias Powalowski wrote: > >> > Just poked around with grub2 and kvm, > >> > you really need the extra space, else it will fail to install. > >> > creating partitions with cfdisk will also leave this space. > >> > >> With nowadays RAID and SSD stuff you want to align your partitions > >> on stripe or blocksizes anyways. Some newer harddisks come with > >> 4Kbyte sectors and have a mapping to 512byte sectors so old OS'es > >> can work with it. You really want to align such devices, as > >> performance is terrible without it. > >> > > > > Err... I'm a bit lost. But anyway, you agree we need the -D flag? > > I'm not really into these alignment things, but if you guys say I > > should apply it, then I'll do it. > > I vote for it :) committed. thank you. http://projects.archlinux.org/aif.git/commit/?h=experimental&id=aa80222dfd8… if anything is broken, blame Mark :) Dieter

1 0

[arch-releng] Note about removing archiso_early hook.
by Gerardo Exequiel Pozzi 05 Mar '10

05 Mar '10

Hi, just to leave in some visible place this experiment if some is interested. My idea: remove archiso_early hook, and take the advantage of symlink /dev/disk/by-label/XYZ created by 60-persistent-storage.rules installed via udev hook, without doing other tricks. Result: * OK: when booting in cdrom mode. * FAIL: when booting in disk mode. The reason is because hybrid mode (via isohybrid). * The LABEL is defined at first sector of iso "sda" (simulate a MBR) also there is a fake partition "sda1" (with offset 1 [+512 bytes ahead of sda]) so you can create another partitions on the destination media. * When mount the UDF filesystem, sda is mounted and not sda1 (sda1 is just a fake part to avoid overwriting data). * There is a udev rule at 60-persistent-storage.rules that make all parent (sda in this case) ID_* env data available to childrens (sda1 in this case), so sda1 inherits the LABEL, and this data is not overwrited because blkid fail (remember sda1 is not a real partition) Consequence: * /dev/disk/by-label/XYZ points to /dev/sda1 instead of /dev/sda. (when isohybrid offset > 0) What works: * if offset = 0 will works, but doing this we lost the capability of create another partitions in the medium that .iso is copied (USB media) * Also works on real partition scheme (usb drive with the real MBR, and real partiton with label) * As said before if boot as cdrom. Conclusion: * Keep with archiso_early, that is pretty simple and does not require other tools [#1] If some is interested in changes for your personal usage, the patch is here [#2] [#1] http://bugs.archlinux.org/task/18523 [mkinitcpio] usb_id, ata_id, scsi_id, path_id: should be added to udev install hook? [#2] https://gist.github.com/1f6f5a67008bf3b927ee -- Gerardo Exequiel Pozzi ( djgera ) http://www.djgera.com.ar KeyID: 0x1B8C330D Key fingerprint = 0CAA D5D4 CD85 4434 A219 76ED 39AB 221B 1B8C 330D

2 3

[arch-releng] automated iso builds
by Dieter Plaetinck 05 Mar '10

05 Mar '10

so i blew the dust of our releng chroots, updated them, installed latest archiso from git and so on. i built this experimental isolinux image (with testing repo enabled!), basically to see if everything still works. http://build.archlinux.org/isos/archlinux-2010.02.09-netinstall-i686.iso through incremental updates, continuous integration and crowd-sourced testing i want to set up automated builds and host all media in the cloud. (if you know what i mean ;-) btw: the image sort of boots, but it can't find /dev/archiso. /me off. to be continued. Dieter

2 21

[arch-releng] [PATCH 5/6] Fix doc path in motd
by Gerardo Exequiel Pozzi 05 Mar '10

05 Mar '10

Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386(a)yahoo.com.ar> --- configs/syslinux-iso/overlay/etc/motd | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/configs/syslinux-iso/overlay/etc/motd b/configs/syslinux-iso/overlay/etc/motd index bbe312f..00e4775 100644 --- a/configs/syslinux-iso/overlay/etc/motd +++ b/configs/syslinux-iso/overlay/etc/motd @@ -1,7 +1,7 @@ ************************************************************** * To begin installation, run /arch/setup * * You can find documentation at * -* /arch/docs/official_installation_guide_en * +* /usr/share/aif/docs/official_installation_guide_en * * * * i18n: Use the 'km' utility to change your keyboard layout * * and console font. * -- 1.7.0.1

1 1

[arch-releng] [PATCH 3/4] Welcome back copytoram= option
by Gerardo Exequiel Pozzi 01 Mar '10

01 Mar '10

Fixes FS#17182: copytoram=y does not work with aufs The solution is simple, just use a directory outside aufs tree. * If copytoram=y, then another tmpfs is mounted but on /tmpfs.sqfs * Add a new option: tmpfs_sqfs_size= (default to 75%) * Lock cdrom drive is not locked if copytoram=y Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386(a)yahoo.com.ar> --- archiso/hooks/archiso | 27 ++++++++++++++++++++++----- archiso/install/archiso | 1 + 2 files changed, 23 insertions(+), 5 deletions(-) diff --git a/archiso/hooks/archiso b/archiso/hooks/archiso index 9997b38..1ac4d86 100644 --- a/archiso/hooks/archiso +++ b/archiso/hooks/archiso @@ -15,6 +15,12 @@ _mnt_squashfs() { img_name="${img_fullname%.*}" tmp_mnt="/tmpfs/mnt/${img_name}" + if [ "${copytoram}" = "y" ]; then + msg ":: Copying squashfs image to RAM" + /bin/cp "${img}" "/tmpfs.sqfs/${img_fullname}" + img="/tmpfs.sqfs/${img_fullname}" + fi + msg "::: Adding new aufs branch: ${img_name}" mkdir -p "${tmp_mnt}" /bin/mount -r -t squashfs "${img}" "${tmp_mnt}" @@ -34,6 +40,10 @@ run_hook () { tmpfs_size="75%" fi + if [ "x${tmpfs_sqfs_size}" = "x" ]; then + tmpfs_sqfs_size="75%" + fi + if [ "x${isomounts}" != "x" ]; then isomounts="/bootmnt/${isomounts}" else @@ -55,6 +65,12 @@ archiso_mount_handler() { mount -t tmpfs -o "size=${tmpfs_size}" tmpfs /tmpfs msg "done." + if [ "${copytoram}" = "y" ]; then + msg -n ":: Mounting tmpfs (for squashfs), size=${tmpfs_sqfs_size}..." + mount -t tmpfs -o "size=${tmpfs_sqfs_size}" tmpfs /tmpfs.sqfs + msg "done." + fi + msg ":: Waiting for boot device..." while ! poll_device ${archisodevice} 30; do echo "ERROR: boot device didn't show up after 30 seconds..." @@ -114,13 +130,14 @@ archiso_mount_handler() { # Bind our bootmnt dir into the live system _mnt_bind /bootmnt "${newroot}/bootmnt" - if [ "${FSTYPE}" = "iso9660" -o "${FSTYPE}" = "udf" ]; then - if [ -d /proc/sys/dev/cdrom ]; then - echo 0 > /proc/sys/dev/cdrom/lock - echo 0 > /proc/sys/dev/cdrom/autoeject + if [ "${copytoram}" != "y" ]; then + if [ "${FSTYPE}" = "iso9660" ] || [ "${FSTYPE}" = "udf" ]; then + if [ -d /proc/sys/dev/cdrom ]; then + echo 0 > /proc/sys/dev/cdrom/lock + echo 0 > /proc/sys/dev/cdrom/autoeject + fi fi fi - } # vim:ft=sh:ts=4:sw=4:et: diff --git a/archiso/install/archiso b/archiso/install/archiso index 4d36412..d01ab25 100644 --- a/archiso/install/archiso +++ b/archiso/install/archiso @@ -10,6 +10,7 @@ install () BINARIES="" FILES="" add_dir /tmpfs + add_dir /tmpfs.sqfs add_dir /bootmnt add_device /dev/loop0 b 7 0 SCRIPT="archiso" -- 1.7.0

1 1