December 2012 - Libguestfs - libguestfs List Archives

[PATCH] sysprep: handle distro specific sysv scripts
by Olaf Hering 12 Dec '13

12 Dec '13

Currently firstboot would only work on redhat-based images. Handle redhat-based, suse-based and debian guests, error out in case of an unknown distro. Update firstboot.sh: - make sure scripts exists and can be executed - add LSB header to avoid insserv warnings later on - run script only if called with "start" Update functions, pass only required options. Signed-off-by: Olaf Hering <olaf(a)aepfle.de> diff --git a/sysprep/firstboot.ml b/sysprep/firstboot.ml index 97cd8a9..c5296a1 100644 --- a/sysprep/firstboot.ml +++ b/sysprep/firstboot.ml @@ -28,14 +28,35 @@ let firstboot_dir = "/usr/lib/virt-sysprep" let firstboot_sh = sprintf "\ #!/bin/sh - +### BEGIN INIT INFO +# Provides: virt-sysprep +# Required-Start: $null +# Should-Start: $all +# Required-Stop: $null +# Should-Stop: $all +# Default-Start: 2 3 5 +# Default-Stop: 0 1 6 +# Short-Description: Start scripts to run once at next boot +# Description: Start scripts to run once at next boot +# These scripts run the first time the guest boots, +# and then are deleted. Output or errors from the scripts +# are written to ~root/virt-sysprep-firstboot.log. +### END INIT INFO + d=%s/scripts logfile=~root/virt-sysprep-firstboot.log -for f in $d/* ; do - echo '=== Running' $f '===' >>$logfile - $f >>$logfile 2>&1 - rm $f -done +if test \"$1\" = \"start\" +then + for f in $d/* ; do + if test -x \"$f\" + then + echo '=== Running' $f '===' >>$logfile + $f >>$logfile 2>&1 + rm -f $f + fi + done +fi " firstboot_dir let firstboot_service = sprintf "\ @@ -56,7 +77,7 @@ WantedBy=default.target let failed fs = ksprintf (fun msg -> failwith (s_"firstboot: failed: " ^ msg)) fs -let rec install_service g root = +let rec install_service g distro = g#mkdir_p firstboot_dir; g#mkdir_p (sprintf "%s/scripts" firstboot_dir); g#write (sprintf "%s/firstboot.sh" firstboot_dir) firstboot_sh; @@ -64,18 +85,18 @@ let rec install_service g root = (* systemd, else assume sysvinit *) if g#is_dir "/etc/systemd" then - install_systemd_service g root + install_systemd_service g else - install_sysvinit_service g root + install_sysvinit_service g distro (* Install the systemd firstboot service, if not installed already. *) -and install_systemd_service g root = +and install_systemd_service g = g#write (sprintf "%s/firstboot.service" firstboot_dir) firstboot_service; g#mkdir_p "/etc/systemd/system/default.target.wants"; g#ln_sf (sprintf "%s/firstboot.service" firstboot_dir) "/etc/systemd/system/default.target.wants" -and install_sysvinit_service g root = +and install_sysvinit_redhat g = g#mkdir_p "/etc/rc.d/rc2.d"; g#mkdir_p "/etc/rc.d/rc3.d"; g#mkdir_p "/etc/rc.d/rc5.d"; @@ -86,12 +107,51 @@ and install_sysvinit_service g root = g#ln_sf (sprintf "%s/firstboot.sh" firstboot_dir) "/etc/rc.d/rc5.d/99virt-sysprep-firstboot" +(* Make firstboot.sh look like a runlevel script to avoid insserv warnings. *) +and install_sysvinit_suse g = + g#mkdir_p "/etc/init.d/rc2.d"; + g#mkdir_p "/etc/init.d/rc3.d"; + g#mkdir_p "/etc/init.d/rc5.d"; + g#ln_sf (sprintf "%s/firstboot.sh" firstboot_dir) + "/etc/init.d/virt-sysprep-firstboot"; + g#ln_sf "../virt-sysprep-firstboot" + "/etc/init.d/rc2.d/S99virt-sysprep-firstboot"; + g#ln_sf "../virt-sysprep-firstboot" + "/etc/init.d/rc3.d/S99virt-sysprep-firstboot"; + g#ln_sf "../virt-sysprep-firstboot" + "/etc/init.d/rc5.d/S99virt-sysprep-firstboot" + +and install_sysvinit_debian g = + g#mkdir_p "/etc/init.d"; + g#mkdir_p "/etc/rc2.d"; + g#mkdir_p "/etc/rc3.d"; + g#mkdir_p "/etc/rc5.d"; + g#ln_sf (sprintf "%s/firstboot.sh" firstboot_dir) + "/etc/init.d/virt-sysprep-firstboot"; + g#ln_sf "/etc/init.d/virt-sysprep-firstboot" + "/etc/rc2.d/S99virt-sysprep-firstboot"; + g#ln_sf "/etc/init.d/virt-sysprep-firstboot" + "/etc/rc3.d/S99virt-sysprep-firstboot"; + g#ln_sf "/etc/init.d/virt-sysprep-firstboot" + "/etc/rc5.d/S99virt-sysprep-firstboot" + +and install_sysvinit_service g distro = + match distro with + | ("fedora"|"rhel"|"centos"|"scientificlinux"|"redhat-based") -> + install_sysvinit_redhat g + | ("opensuse"|"sles"|"suse-based") -> + install_sysvinit_suse g + | "debian" -> + install_sysvinit_debian g + | _ -> + failed "guest type %s is not supported" distro + let add_firstboot_script g root id content = let typ = g#inspect_get_type root in let distro = g#inspect_get_distro root in match typ, distro with | "linux", _ -> - install_service g root; + install_service g distro; let t = Int64.of_float (Unix.time ()) in let r = string_random8 () in let filename = sprintf "%s/scripts/%Ld-%s-%s" firstboot_dir t r id in -- 1.7.12

2 8

[PATCH 0/2] Add support for rsync.
by Richard W.M. Jones 19 Feb '13

19 Feb '13

An experimental series that adds support for rsync. Rich.

2 5

err from guestfs_launch
by Skippy VonDrake 02 Jan '13

02 Jan '13

This error is with libguestfs 1.20. Calling guestfs_launch() with tracing on: supermin helper [00065ms] finished creating appliance libguestfs: checksum of existing appliance: 2bd6811426e14b6fb661b4d52cebd6c21a41df1a068e8acf6a7a08e3b3bd4ada libguestfs: trace: get_cachedir libguestfs: trace: get_cachedir = "/var/tmp" libguestfs: [00071ms] begin building supermin appliance libguestfs: trace: get_cachedir libguestfs: trace: get_cachedir = "/var/tmp" libguestfs: [00071ms] run febootstrap-supermin-helper libguestfs: command: run: febootstrap-supermin-helper libguestfs: command: run: \ --verbose libguestfs: command: run: \ --copy-kernel libguestfs: command: run: \ -f ext2 libguestfs: command: run: \ /usr/local/lib/guestfs/supermin.d libguestfs: command: run: \ x86_64 libguestfs: command: run: \ /var/tmp/guestfs.lQs0RL/kernel libguestfs: command: run: \ /var/tmp/guestfs.lQs0RL/initrd libguestfs: command: run: \ /var/tmp/guestfs.lQs0RL/root febootstrap-supermin-helper: unrecognized option '--copy-kernel' febootstrap-supermin-helper: build the supermin appliance on the fly Usage: febootstrap-supermin-helper [-options] inputs [...] host_cpu kernel initrd <snip> Same code works on a different system. Both systems are Ubuntu 12.04. However the system that works is using Linux 3.6.4-pv whereas the system that fails is Linux 3.5.4-pv. Do I need to upgrade the kernel?

1 1

Happy holidays
by Richard W.M. Jones 24 Dec '12

24 Dec '12

Happy holidays to all our users. Matt is on holiday this week. I'm on holiday until * January 16th *. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-p2v converts physical machines to virtual machines. Boot with a live CD or over the network (PXE) and turn machines into KVM guests. http://libguestfs.org/virt-v2v

1 0

febootstrap and download packages
by Evaggelos Balaskas 22 Dec '12

22 Dec '12

Something that i would like to have, is the ability to keep the download packages in a temp directory when using febootstrap. I understand that using ferora/archlinux (distros with a lot of package updates every day) can increase the size of this temp directory rapidly and at some point will have a lot of deprecate packages. But the daily use of febootstrap (or building libguestfs) makes your network suffer for downloading the same packages again and again. For build libguestfs on archlinux is an amount of 140Mb. On my machine i've found two solutions: a. creating a custom local repo and move it to the top of my pacman.conf, so it would access this local repo without the need of internet b. hacking a febootstrap_pacman.ml to use /tmp/cachedir (or a dir on my home). Do you believe that a parameter like "--save-temp" eg. "--save-pkgs" could matter ? -- Evaggelos Balaskas - Unix System Engineer http://gr.linkedin.com/in/evaggelosbalaskas

3 16

Supporting btrfs subvolumes during inspection
by Matthew Booth 21 Dec '12

21 Dec '12

We've currently got a bug in libguestfs which means we can't inspect filesystems in btrfs subvolumes: https://bugzilla.redhat.com/show_bug.cgi?id=824021 This is the default configuration if you select btrfs in F17+. The issue is that it requires an api to fix it, as the return values of inspect_os, inspect_get_filesystems and inspect_get_mountpoints can't express a btrfs subvolume as they're expected to be the names of block devices. As a starter for 10, I propose the addition of parallel apis suffixed _ext which return annotated device descriptions, e.g.: block:/dev/vda2 btrfsvol:/dev/vda4,root The string before the colon is a descriptor. The format of the string after the colon is custom to the descriptor, and is described in the API. A client program can ignore strings with unknown descriptors. New descriptors can be added as required. The old apis would return the same as the _ext versions, except they would only return block devices. Other devices would be omitted. Matt -- Matthew Booth, RHCA, RHCSS Red Hat Engineering, Virtualisation Team GPG ID: D33C3490 GPG FPR: 3733 612D 2D05 5458 8A8A 1600 3441 EA19 D33C 3490

2 7

Can't access remote URI
by Wanlong Gao 20 Dec '12

20 Dec '12

Hi Rich, We just found that the libguestfs can't access the remote URI. When doing guestfs__add_drive_opts(), we always add files from local system, it's related the -c|--connect option. As I know, we are using local kernel to lunch the min-guest, and it's hard to attach remote disks to our local min-guest. Our test team found this problem by using following command, # virt-sysprep -c qemy+ssh://<host>/system -d domname Then, for example the path of remote disk is /work/rhel.img, but we are about to access the /work/rhel.img locally. So, IMHO, if we are about to not support the remote URI, we should give a error message first. But access local disks instead of remote disks are definitely wrong here. Maybe we also need document this. Any thoughts? Thanks, Wanlong Gao

4 8

Need DETAILED help using p2v
by Bob Buckley 19 Dec '12

19 Dec '12

This is something I want to do over my holiday vacation to continue to broaden my skills. I am experimenting (learning) about moving a Win7 physical drive to a VM under my CSB. I'm struggling a little on the steps when using our RH documentation (perhaps it is clear if you are an RHCE but it isn't to someone trying to learn from it initially combined with Google searches) so asking if anyone has any one-off how to scripts for this scenario? The two drives (Win7 and CSB) are two different hard drives both bootable on my laptop so it should be simple but... I have been pointed to several existing pages that are all written for the EXPERIENCED RHCE but not for the novice. Appreciate any assistance you can provide. Thanks, Bob Bob "B 2 " Buckley Sr. Storage Solutions Architect, Intelligence & National Security 719.330.9329 (Mobile), 571.421.8160 (Office) bbuckley(a)redhat.com

2 1

Re: [Libguestfs] inspect_os error with Win7 image
by Richard W.M. Jones 18 Dec '12

18 Dec '12

On Mon, Dec 17, 2012 at 04:46:02PM -0500, Skippy VonDrake wrote: > > As far as we know, this means the registry is really corrupt (or > > something odd has happened, such as it's been zero-padded). > > It appears that the registry did have something wrong with it. > > I booted the image in Xen and saw a message window pop up: > "Registry Recovery - One of the files containing the system's > Registry data had to be recovered by use of a log or alternate copy > The recovery was successful." > > And after I shut the VM back down, guestfs was able to load > and continue past the call to guestfs_inspect_os() with no > problems. > > Is there a way I can get more data to determine exactly where > the hivex calls are failing? Maybe then a fix could be made to > still return the root devices even when the registry is a little > "off". It's possible we could add a HIVEX_OPEN_FIXUP flag or something like that. In this case (empty hbins) it would probably be safe, but I'd worry about trying to "fix" more egregious registry errors, particularly since we don't fully understand the format. > With this in mind I altered 'hivex.c', function do_hivex_open(), > by commenting out the code around the statements: > flags |= HIVEX_OPEN_VERBOSE; > flags |= HIVEX_OPEN_DEBUG; > Effectively hard-coding them. Hoping that more info would be > generated. However that obviously wasn't enough for the > process just seemed to "hang". You'll also need to enable libguestfs verbose messages: guestfs_set_verbose (g, 1); That should dump everything to stderr, and with hivex debug you really will get a very large amount of debugging info. If it hangs again, let us know where (please file a bug with all the verbose output until the hang). > What else would need changing to get more details generated > from guestfs_inspect_os->hivex_open? If you were feeling very adventurous, you could run the "visualizer" program from: https://github.com/libguestfs/hivex/tree/master/lib/tools That's the program I used when I was reverse-engineering the registry format, and the source of the program has lots of useful information about the format. However I don't think it will tell you anything very interesting in this case. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones libguestfs lets you edit virtual machines. Supports shell scripting, bindings from many languages. http://libguestfs.org

2 4

Issues with GObject bindings
by Richard W.M. Jones 18 Dec '12

18 Dec '12

Matt, further to the conversation today. When you run the tests (make -C gobject check) you'll find it now prints the descriptive message: "GObject bind tests skipped: Int64, OInt64 and OStringList are all broken" That nicely sums up the first two problems. (1) When I tested passing the full range of int64, it turns out to be broken. There are probably two layers of brokenness, ie. Javascript itself has broken integers as a design failure of the language, but also I convinced myself at some point that something in GObject truncates all integers to 32 bits. BTW also check it works on a 32 bit architecture. (2) OStringList: IIRC I added this type to the generator but was unable / didn't have time to add it to the GObject bindings. (3) Stability of the API: GObject doesn't understand the once_had_no_optargs feature of the generator, which means that converting a no-optargs call to an optargs call (which is now permitted) will break the GObject API. You may or may not care about this. (4) Coverity found a potential bug in header inclusion. It seems that GObject headers recursively include themselves. Both Ondrej and I tried different approaches to fixing this but were unable to make it work. (5) I heavily refactored the code in generator/gobject.ml. Although I checked that the output of the generator didn't change, you might also want to check those commits for sanity. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://et.redhat.com/~rjones/virt-top

1 0