Bug#411814: qemu: please, reopen bug #411814
Gonzalo Pérez de Olaguer Córdoba
Version: 0.10.6-1
Followup-For: Bug #411814
I'm asking to reopen this bug because the symptoms are quite the same.
I'm using the version of kqemu and dependencies from squeeze, although
the base system is lenny.
Trying to install debian into a qemu image using the lenny netinst disk
image some parts of the installation (disk formatting and
package installation) go veru slow, and eventually dumps core (segmentation
fault). This only happens when using qcow images.
When using images in raw format everything goes fast.
I suspect this is a problem related to disk write access in qcow format.
For your information, I'm creating images this way:
qemu-img -f qcow lenny-base.qcow 10G
and running qemu this way:
sudo qemu -m 256 -net nic,vlan=1,macaddr=52:54:00:12:34:51 -net tap,vlan=1 -cdrom debian-502a-i386-netinst.iso -boot d lenny-base.qcow
I have tried also debian-501-i386-netinst.iso
I have tried also qemu version 0.10.4-1 with same results.
Having 1G memory, 2G swap and at least 20G free disk space.
-- System Information:
Debian Release: 5.0.2
APT prefers stable
APT policy: (500, 'stable')
Architecture: i386 (i686)
Kernel: Linux 2.6.26-2-686 (SMP w/2 CPU cores)
Locale: LANG=es_ES.UTF-8, LC_CTYPE=es_ES.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash
Versions of packages qemu depends on:
ii bochsbios 2.3.7+20090416-1 BIOS for the Bochs emulator
ii libasound2 1.0.20-2 shared library for ALSA applicatio
ii libbluetooth3 4.42-2 Library to use the BlueZ Linux Blu
ii libc6 2.7-18 GNU C Library: Shared libraries
ii libesd0 0.2.36-3 Enlightened Sound Daemon - Shared
ii libgnutls26 2.6.6-1 the GNU TLS library - runtime libr
ii libncurses5 5.7+20081213-1 shared libraries for terminal hand
ii libpulse0 0.9.15-2 PulseAudio client libraries
ii libsdl1.2debian 1.2.13-2 Simple DirectMedia Layer
ii libvdeplug2 2.2.2-3 Virtual Distributed Ethernet - Plu
ii libx11-6 2:1.1.5-2 X11 client-side library
ii openbios-ppc 1.0-3 PowerPC Open Firmware
ii openbios-sparc 1.0-1 SPARC Open Firmware
ii openhackware 0.4.1-4 OpenFirmware emulator for PowerPC
ii vgabios 0.6c-1 VGA BIOS software for the Bochs an
ii zlib1g 1:1.2.3.3.dfsg-12 compression library - runtime
Versions of packages qemu recommends:
ii debootstrap 1.0.10lenny1 Bootstrap a basic Debian system
ii sharutils 1:4.6.3-1 shar, unshar, uuencode, uudecode
ii vde2 2.2.2-3 Virtual Distributed Ethernet
Versions of packages qemu suggests:
ii kqemu-source 1.4.0~pre1-2 Source for the QEMU Accelerator mo
ii samba 2:3.2.5-4lenny6 a LanManager-like file and printer
ii sudo 1.6.9p17-2 Provide limited super user privile
-- no debconf information
--
To UNSUBSCRIBE, email to debian-bugs-...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listm...@lists.debian.org
Aurelien Jarno
> Package: qemu
> Version: 0.10.6-1
> Followup-For: Bug #411814
>
>
> I'm asking to reopen this bug because the symptoms are quite the same.
>
> I'm using the version of kqemu and dependencies from squeeze, although
> the base system is lenny.
>
> Trying to install debian into a qemu image using the lenny netinst disk
> image some parts of the installation (disk formatting and
> package installation) go veru slow, and eventually dumps core (segmentation
> fault). This only happens when using qcow images.
>
> When using images in raw format everything goes fast.
>
> I suspect this is a problem related to disk write access in qcow format.
>
> For your information, I'm creating images this way:
>
> qemu-img -f qcow lenny-base.qcow 10G
>
> and running qemu this way:
>
> sudo qemu -m 256 -net nic,vlan=1,macaddr=52:54:00:12:34:51 -net tap,vlan=1 -cdrom debian-502a-i386-netinst.iso -boot d lenny-base.qcow
>
I am almost sure it's a different problem. I guess your problems are due
to cache policy as shown in the NEWS.Debian file (see below).
You should try using '-drive file=lenny-base.qcow,cache=writeback'
instead of simply 'lenny-base.qcow'.
Aurelien
qemu (0.10.3-2) unstable; urgency=low
Starting with QEMU 0.10.0, it is possible to control how the host
cache is used to access block data, using the cache= suboption of the
-drive option. The following suboptions are available:
* none: The host page cache is entirely avoided.
* writeback (default in QEMU 0.9.x): Writeback caching reports data
writes as completed as soon as the data is present in the host page
cache. This is safe as long as you trust your host. If your host
crashes or loses power, then the guest may experience data
corruption.
* writethrough (default in QEMU 0.10.x): The host page cache is used
to read and write data but write notification is sent to the guest
only when the data has been reported as written by the storage
subsystem.
Note that depending on your configuration (filesystem, encryption,
kernel version, etc.), disk accesses can be very slow with the default
cache policy (writethrough). You can use the writeback cache policy
instead, but the data integrity is not assured anymore.
See qemu(1) for more details.
-- Aurelien Jarno <aur...@debian.org> Sun, 03 May 2009 23:22:29 +0200
--
Aurelien Jarno GPG: 1024D/F1BCDB73
aure...@aurel32.net http://www.aurel32.net
Gonzalo Pérez de Olaguer Córdoba
el Tue, 01 Sep 2009 17:57:28 +0200 escribiste:
> I am almost sure it's a different problem. I guess your problems are due
> to cache policy as shown in the NEWS.Debian file (see below).
>
> You should try using '-drive file=lenny-base.qcow,cache=writeback'
> instead of simply 'lenny-base.qcow'.
Yes, that's it. Using qcow2 format also solves the problem.
Thanks a lot. My apologies for the inconvenience, I should have read
the documentation again after upgrade.
--
Gonzalo Pérez de Olaguer Córdoba <gp...@iies.es> --- www.gpoc.es
PGP key 2861C704 --- F206 5671 6789 425D 111C 1302 214F 1934 2861 C704