[ironic] Each deploy overwrites the entire target disk
Dears,
I have a question regarding the default behaviour of Ironic Python Agent - it seems that each deploy causes the entire target disk to be overwritten. I have noticed this after replacing disks with larger ones, the deployment process started taking considerably longer and I observed that each sector is rewritten on the target device. Is that expected? Can I somehow avoid it? I am using whole-disk, qcow2 images that are much smaller than the disk size (even in raw form). I am running Ironic Wallaby. The deployment method is direct.
Kind regards, -yoctozepto
Hi!
On Mon, Feb 7, 2022 at 4:50 PM Radosław Piliszek < radoslaw.piliszek@gmail.com> wrote:
Dears,
I have a question regarding the default behaviour of Ironic Python Agent - it seems that each deploy causes the entire target disk to be overwritten. I have noticed this after replacing disks with larger ones, the deployment process started taking considerably longer and I observed that each sector is rewritten on the target device. Is that expected? Can I somehow avoid it?
I think we only overwrite metadata on the target disk, not zero it. Are you sure it's not cleaning?
Dmitry
I am using whole-disk, qcow2 images that are much smaller than the disk size (even in raw form). I am running Ironic Wallaby. The deployment method is direct.
Kind regards, -yoctozepto
On Tue, 8 Feb 2022 at 09:15, Dmitry Tantsur dtantsur@redhat.com wrote:
Hi!
On Mon, Feb 7, 2022 at 4:50 PM Radosław Piliszek radoslaw.piliszek@gmail.com wrote:
Dears,
I have a question regarding the default behaviour of Ironic Python Agent - it seems that each deploy causes the entire target disk to be overwritten. I have noticed this after replacing disks with larger ones, the deployment process started taking considerably longer and I observed that each sector is rewritten on the target device. Is that expected? Can I somehow avoid it?
I think we only overwrite metadata on the target disk, not zero it. Are you sure it's not cleaning?
Nope, the cleaning process is quick, this is on deploy. The command that does the entire overwrite is qemu-img convert from IPA. I wonder if some semantics changed in the meantime. I am using Debian Buster (10) based images. So with this https://packages.debian.org/buster/qemu-utils
-yoctozepto
On Tue, Feb 8, 2022 at 9:29 AM Radosław Piliszek < radoslaw.piliszek@gmail.com> wrote:
On Tue, 8 Feb 2022 at 09:15, Dmitry Tantsur dtantsur@redhat.com wrote:
Hi!
On Mon, Feb 7, 2022 at 4:50 PM Radosław Piliszek <
radoslaw.piliszek@gmail.com> wrote:
Dears,
I have a question regarding the default behaviour of Ironic Python Agent - it seems that each deploy causes the entire target disk to be overwritten. I have noticed this after replacing disks with larger ones, the deployment process started taking considerably longer and I observed that each sector is rewritten on the target device. Is that expected? Can I somehow avoid it?
I think we only overwrite metadata on the target disk, not zero it. Are
you sure it's not cleaning?
Nope, the cleaning process is quick, this is on deploy. The command that does the entire overwrite is qemu-img convert from IPA. I wonder if some semantics changed in the meantime. I am using Debian Buster (10) based images. So with this https://packages.debian.org/buster/qemu-utils
Ah, you may need this change then: https://review.opendev.org/c/openstack/ironic-lib/+/808993
Dmitry
-yoctozepto
On Tue, 8 Feb 2022 at 11:26, Dmitry Tantsur dtantsur@redhat.com wrote:
On Tue, Feb 8, 2022 at 9:29 AM Radosław Piliszek radoslaw.piliszek@gmail.com wrote:
On Tue, 8 Feb 2022 at 09:15, Dmitry Tantsur dtantsur@redhat.com wrote:
Hi!
On Mon, Feb 7, 2022 at 4:50 PM Radosław Piliszek radoslaw.piliszek@gmail.com wrote:
Dears,
I have a question regarding the default behaviour of Ironic Python Agent - it seems that each deploy causes the entire target disk to be overwritten. I have noticed this after replacing disks with larger ones, the deployment process started taking considerably longer and I observed that each sector is rewritten on the target device. Is that expected? Can I somehow avoid it?
I think we only overwrite metadata on the target disk, not zero it. Are you sure it's not cleaning?
Nope, the cleaning process is quick, this is on deploy. The command that does the entire overwrite is qemu-img convert from IPA. I wonder if some semantics changed in the meantime. I am using Debian Buster (10) based images. So with this https://packages.debian.org/buster/qemu-utils
Ah, you may need this change then: https://review.opendev.org/c/openstack/ironic-lib/+/808993
Thanks, that's it!
-yoctozepto
Dmitry
-yoctozepto
-- Red Hat GmbH, https://de.redhat.com/ , Registered seat: Grasbrunn, Commercial register: Amtsgericht Muenchen, HRB 153243, Managing Directors: Charles Cachera, Brian Klemm, Laurie Krebs, Michael O'Neill
participants (2)
-
Dmitry Tantsur
-
Radosław Piliszek