Re: [one-users] OpenNebula ceph snapshots problem.
Hi Leszek, using Ceph and qcow2 is not something we have leveraged. Maybe someone from the community can report their experiences? I've created a feature report to see if we can add support for this. http://dev.opennebula.org/issues/2947 Regards, Jaime On Thu, May 15, 2014 at 9:32 AM, Leszek Master keks...@gmail.com wrote: I can't migrate to NFS because the disk io performance won't handle my VM's. There isn't any sunstone modification to use rbd v2 images snapshots ? I can create them using rbd snap create, but don't have the possibility to use it via sunstone. 2014-05-14 16:31 GMT+02:00 Vladislav Gorbunov vadi...@gmail.com: Can you write me what i need to do to migrate to qcow2 format? I've got running vm's that cannot be deleted. I tried migrating them form rbd v1 to v2 but this doesn't solved my problem. You need NFS datastore. Only NFS or SSH datastore support qcow2 file format. But Opennebula can't move image between datastores. 1. Shutdown vm 2. Convert from ceph to qcow2 format: export image=vm-disk-1; qemu-img convert -O qcow2 rbd:`oneimage show $image|awk '/^SOURCE/ {print $3}'` /var/lib/one/datastore/123/$image.qcow2 3. Rename ceph image: oneimage rename $image $image.bak 4. Import qcow2 image to NFS datastore with id 123 for example: oneimage create -d nfsdatastorename --name $image --type os --persistent --source /var/lib/one/datastores/123/$image.qcow2 --driver qcow2 --size `rbd info rados/$image|awk '/size/ {print $5*4}'` 5. Start vm 2014-05-14 22:08 GMT+12:00 Leszek Master keks...@gmail.com: Can you write me what i need to do to migrate to qcow2 format? I've got running vm's that cannot be deleted. I tried migrating them form rbd v1 to v2 but this doesn't solved my problem. 2014-04-30 12:41 GMT+02:00 Javier Fontan jfon...@opennebula.org: I'm not sure I follow you with that. Just changing the TM from ceph to qcow2 is something that probably doesn't work. One of the things that can be failing is that getting the free space from ceph and qcow2 is totally different and it may be that the datastore is reporting 0 bytes free. In this case the scheduler will never start a VM that uses that datastore. On Wed, Apr 23, 2014 at 9:18 AM, Leszek Master keks...@gmail.com wrote: I tried to change rbd image version from 1 to 2 but it doesn't work. But when i changed in my ceph datastore TM_MAD to qcow2 the new VM instantized from image that was in the datastore stuck on PENDING. Is there any way to get this working with existing images? 2014-04-21 14:53 GMT+02:00 Stuart Longland stua...@vrt.com.au: On 17/04/14 01:18, Leszek Master wrote: 1) I'm using OpenNebula 4.4 with ceph datastore. When i try to make snapshot i've got error: 3) Is there any way to use copy-on-write in opennebula 4.4 ? I did some experimental work on this with our OpenNebula 4.4 instance. It relies on modern kernel's ability to map version-2 RBD images using the kernel rbd driver, so it needs kernel 3.10 (? Not sure of exact version; we're using 3.13). This newer format allows copy-on-write cloning. Hot snapshots work, however I haven't managed to trigger a deferred snapshot, so that's untested. The driver also makes use of hypervisor-side cache using FlashCache, allocating slices of a nominated LVM volume to boost performance. I've thrown my work into a git repository for now, consider this very much pre-alpha. If you're brave, feel free to give it a go: git clone git:// git.longlandclan.yi.org/opennebula-ceph-flashcache.git Regards, -- Stuart Longland Systems Engineer _ ___ \ /|_) | T: +61 7 3535 9619 \/ | \ | 38b Douglas StreetF: +61 7 3535 9699 SYSTEMSMilton QLD 4064 http://www.vrt.com.au ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org -- Javier Fontán Muiños Developer OpenNebula - The Open Source Toolkit for Data Center Virtualization www.OpenNebula.org | @OpenNebula | github.com/jfontan ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org -- Jaime Melis Project Engineer OpenNebula - Flexible Enterprise Cloud Made Simple www.OpenNebula.org | jme...@opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] OpenNebula ceph snapshots problem.
I can't migrate to NFS because the disk io performance won't handle my VM's. There isn't any sunstone modification to use rbd v2 images snapshots ? I can create them using rbd snap create, but don't have the possibility to use it via sunstone. 2014-05-14 16:31 GMT+02:00 Vladislav Gorbunov vadi...@gmail.com: Can you write me what i need to do to migrate to qcow2 format? I've got running vm's that cannot be deleted. I tried migrating them form rbd v1 to v2 but this doesn't solved my problem. You need NFS datastore. Only NFS or SSH datastore support qcow2 file format. But Opennebula can't move image between datastores. 1. Shutdown vm 2. Convert from ceph to qcow2 format: export image=vm-disk-1; qemu-img convert -O qcow2 rbd:`oneimage show $image|awk '/^SOURCE/ {print $3}'` /var/lib/one/datastore/123/$image.qcow2 3. Rename ceph image: oneimage rename $image $image.bak 4. Import qcow2 image to NFS datastore with id 123 for example: oneimage create -d nfsdatastorename --name $image --type os --persistent --source /var/lib/one/datastores/123/$image.qcow2 --driver qcow2 --size `rbd info rados/$image|awk '/size/ {print $5*4}'` 5. Start vm 2014-05-14 22:08 GMT+12:00 Leszek Master keks...@gmail.com: Can you write me what i need to do to migrate to qcow2 format? I've got running vm's that cannot be deleted. I tried migrating them form rbd v1 to v2 but this doesn't solved my problem. 2014-04-30 12:41 GMT+02:00 Javier Fontan jfon...@opennebula.org: I'm not sure I follow you with that. Just changing the TM from ceph to qcow2 is something that probably doesn't work. One of the things that can be failing is that getting the free space from ceph and qcow2 is totally different and it may be that the datastore is reporting 0 bytes free. In this case the scheduler will never start a VM that uses that datastore. On Wed, Apr 23, 2014 at 9:18 AM, Leszek Master keks...@gmail.com wrote: I tried to change rbd image version from 1 to 2 but it doesn't work. But when i changed in my ceph datastore TM_MAD to qcow2 the new VM instantized from image that was in the datastore stuck on PENDING. Is there any way to get this working with existing images? 2014-04-21 14:53 GMT+02:00 Stuart Longland stua...@vrt.com.au: On 17/04/14 01:18, Leszek Master wrote: 1) I'm using OpenNebula 4.4 with ceph datastore. When i try to make snapshot i've got error: 3) Is there any way to use copy-on-write in opennebula 4.4 ? I did some experimental work on this with our OpenNebula 4.4 instance. It relies on modern kernel's ability to map version-2 RBD images using the kernel rbd driver, so it needs kernel 3.10 (? Not sure of exact version; we're using 3.13). This newer format allows copy-on-write cloning. Hot snapshots work, however I haven't managed to trigger a deferred snapshot, so that's untested. The driver also makes use of hypervisor-side cache using FlashCache, allocating slices of a nominated LVM volume to boost performance. I've thrown my work into a git repository for now, consider this very much pre-alpha. If you're brave, feel free to give it a go: git clone git:// git.longlandclan.yi.org/opennebula-ceph-flashcache.git Regards, -- Stuart Longland Systems Engineer _ ___ \ /|_) | T: +61 7 3535 9619 \/ | \ | 38b Douglas StreetF: +61 7 3535 9699 SYSTEMSMilton QLD 4064 http://www.vrt.com.au ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org -- Javier Fontán Muiños Developer OpenNebula - The Open Source Toolkit for Data Center Virtualization www.OpenNebula.org | @OpenNebula | github.com/jfontan ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] OpenNebula ceph snapshots problem.
Can you write me what i need to do to migrate to qcow2 format? I've got running vm's that cannot be deleted. I tried migrating them form rbd v1 to v2 but this doesn't solved my problem. 2014-04-30 12:41 GMT+02:00 Javier Fontan jfon...@opennebula.org: I'm not sure I follow you with that. Just changing the TM from ceph to qcow2 is something that probably doesn't work. One of the things that can be failing is that getting the free space from ceph and qcow2 is totally different and it may be that the datastore is reporting 0 bytes free. In this case the scheduler will never start a VM that uses that datastore. On Wed, Apr 23, 2014 at 9:18 AM, Leszek Master keks...@gmail.com wrote: I tried to change rbd image version from 1 to 2 but it doesn't work. But when i changed in my ceph datastore TM_MAD to qcow2 the new VM instantized from image that was in the datastore stuck on PENDING. Is there any way to get this working with existing images? 2014-04-21 14:53 GMT+02:00 Stuart Longland stua...@vrt.com.au: On 17/04/14 01:18, Leszek Master wrote: 1) I'm using OpenNebula 4.4 with ceph datastore. When i try to make snapshot i've got error: 3) Is there any way to use copy-on-write in opennebula 4.4 ? I did some experimental work on this with our OpenNebula 4.4 instance. It relies on modern kernel's ability to map version-2 RBD images using the kernel rbd driver, so it needs kernel 3.10 (? Not sure of exact version; we're using 3.13). This newer format allows copy-on-write cloning. Hot snapshots work, however I haven't managed to trigger a deferred snapshot, so that's untested. The driver also makes use of hypervisor-side cache using FlashCache, allocating slices of a nominated LVM volume to boost performance. I've thrown my work into a git repository for now, consider this very much pre-alpha. If you're brave, feel free to give it a go: git clone git://git.longlandclan.yi.org/opennebula-ceph-flashcache.git Regards, -- Stuart Longland Systems Engineer _ ___ \ /|_) | T: +61 7 3535 9619 \/ | \ | 38b Douglas StreetF: +61 7 3535 9699 SYSTEMSMilton QLD 4064 http://www.vrt.com.au ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org -- Javier Fontán Muiños Developer OpenNebula - The Open Source Toolkit for Data Center Virtualization www.OpenNebula.org | @OpenNebula | github.com/jfontan ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] OpenNebula ceph snapshots problem.
I've forgot to tell that i've upgraded my opennebula to 4.6 version. 2014-05-14 12:08 GMT+02:00 Leszek Master keks...@gmail.com: Can you write me what i need to do to migrate to qcow2 format? I've got running vm's that cannot be deleted. I tried migrating them form rbd v1 to v2 but this doesn't solved my problem. 2014-04-30 12:41 GMT+02:00 Javier Fontan jfon...@opennebula.org: I'm not sure I follow you with that. Just changing the TM from ceph to qcow2 is something that probably doesn't work. One of the things that can be failing is that getting the free space from ceph and qcow2 is totally different and it may be that the datastore is reporting 0 bytes free. In this case the scheduler will never start a VM that uses that datastore. On Wed, Apr 23, 2014 at 9:18 AM, Leszek Master keks...@gmail.com wrote: I tried to change rbd image version from 1 to 2 but it doesn't work. But when i changed in my ceph datastore TM_MAD to qcow2 the new VM instantized from image that was in the datastore stuck on PENDING. Is there any way to get this working with existing images? 2014-04-21 14:53 GMT+02:00 Stuart Longland stua...@vrt.com.au: On 17/04/14 01:18, Leszek Master wrote: 1) I'm using OpenNebula 4.4 with ceph datastore. When i try to make snapshot i've got error: 3) Is there any way to use copy-on-write in opennebula 4.4 ? I did some experimental work on this with our OpenNebula 4.4 instance. It relies on modern kernel's ability to map version-2 RBD images using the kernel rbd driver, so it needs kernel 3.10 (? Not sure of exact version; we're using 3.13). This newer format allows copy-on-write cloning. Hot snapshots work, however I haven't managed to trigger a deferred snapshot, so that's untested. The driver also makes use of hypervisor-side cache using FlashCache, allocating slices of a nominated LVM volume to boost performance. I've thrown my work into a git repository for now, consider this very much pre-alpha. If you're brave, feel free to give it a go: git clone git://git.longlandclan.yi.org/opennebula-ceph-flashcache.git Regards, -- Stuart Longland Systems Engineer _ ___ \ /|_) | T: +61 7 3535 9619 \/ | \ | 38b Douglas StreetF: +61 7 3535 9699 SYSTEMSMilton QLD 4064 http://www.vrt.com.au ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org -- Javier Fontán Muiños Developer OpenNebula - The Open Source Toolkit for Data Center Virtualization www.OpenNebula.org | @OpenNebula | github.com/jfontan ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] OpenNebula ceph snapshots problem.
Can you write me what i need to do to migrate to qcow2 format? I've got running vm's that cannot be deleted. I tried migrating them form rbd v1 to v2 but this doesn't solved my problem. You need NFS datastore. Only NFS or SSH datastore support qcow2 file format. But Opennebula can't move image between datastores. 1. Shutdown vm 2. Convert from ceph to qcow2 format: export image=vm-disk-1; qemu-img convert -O qcow2 rbd:`oneimage show $image|awk '/^SOURCE/ {print $3}'` /var/lib/one/datastore/123/$image.qcow2 3. Rename ceph image: oneimage rename $image $image.bak 4. Import qcow2 image to NFS datastore with id 123 for example: oneimage create -d nfsdatastorename --name $image --type os --persistent --source /var/lib/one/datastores/123/$image.qcow2 --driver qcow2 --size `rbd info rados/$image|awk '/size/ {print $5*4}'` 5. Start vm 2014-05-14 22:08 GMT+12:00 Leszek Master keks...@gmail.com: Can you write me what i need to do to migrate to qcow2 format? I've got running vm's that cannot be deleted. I tried migrating them form rbd v1 to v2 but this doesn't solved my problem. 2014-04-30 12:41 GMT+02:00 Javier Fontan jfon...@opennebula.org: I'm not sure I follow you with that. Just changing the TM from ceph to qcow2 is something that probably doesn't work. One of the things that can be failing is that getting the free space from ceph and qcow2 is totally different and it may be that the datastore is reporting 0 bytes free. In this case the scheduler will never start a VM that uses that datastore. On Wed, Apr 23, 2014 at 9:18 AM, Leszek Master keks...@gmail.com wrote: I tried to change rbd image version from 1 to 2 but it doesn't work. But when i changed in my ceph datastore TM_MAD to qcow2 the new VM instantized from image that was in the datastore stuck on PENDING. Is there any way to get this working with existing images? 2014-04-21 14:53 GMT+02:00 Stuart Longland stua...@vrt.com.au: On 17/04/14 01:18, Leszek Master wrote: 1) I'm using OpenNebula 4.4 with ceph datastore. When i try to make snapshot i've got error: 3) Is there any way to use copy-on-write in opennebula 4.4 ? I did some experimental work on this with our OpenNebula 4.4 instance. It relies on modern kernel's ability to map version-2 RBD images using the kernel rbd driver, so it needs kernel 3.10 (? Not sure of exact version; we're using 3.13). This newer format allows copy-on-write cloning. Hot snapshots work, however I haven't managed to trigger a deferred snapshot, so that's untested. The driver also makes use of hypervisor-side cache using FlashCache, allocating slices of a nominated LVM volume to boost performance. I've thrown my work into a git repository for now, consider this very much pre-alpha. If you're brave, feel free to give it a go: git clone git://git.longlandclan.yi.org/opennebula-ceph-flashcache.git Regards, -- Stuart Longland Systems Engineer _ ___ \ /|_) | T: +61 7 3535 9619 \/ | \ | 38b Douglas StreetF: +61 7 3535 9699 SYSTEMSMilton QLD 4064 http://www.vrt.com.au ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org -- Javier Fontán Muiños Developer OpenNebula - The Open Source Toolkit for Data Center Virtualization www.OpenNebula.org | @OpenNebula | github.com/jfontan ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] OpenNebula ceph snapshots problem.
I tried to change rbd image version from 1 to 2 but it doesn't work. But when i changed in my ceph datastore TM_MAD to qcow2 the new VM instantized from image that was in the datastore stuck on PENDING. Is there any way to get this working with existing images? 2014-04-21 14:53 GMT+02:00 Stuart Longland stua...@vrt.com.au: On 17/04/14 01:18, Leszek Master wrote: 1) I'm using OpenNebula 4.4 with ceph datastore. When i try to make snapshot i've got error: 3) Is there any way to use copy-on-write in opennebula 4.4 ? I did some experimental work on this with our OpenNebula 4.4 instance. It relies on modern kernel's ability to map version-2 RBD images using the kernel rbd driver, so it needs kernel 3.10 (? Not sure of exact version; we're using 3.13). This newer format allows copy-on-write cloning. Hot snapshots work, however I haven't managed to trigger a deferred snapshot, so that's untested. The driver also makes use of hypervisor-side cache using FlashCache, allocating slices of a nominated LVM volume to boost performance. I've thrown my work into a git repository for now, consider this very much pre-alpha. If you're brave, feel free to give it a go: git clone git://git.longlandclan.yi.org/opennebula-ceph-flashcache.git Regards, -- Stuart Longland Systems Engineer _ ___ \ /|_) | T: +61 7 3535 9619 \/ | \ | 38b Douglas StreetF: +61 7 3535 9699 SYSTEMSMilton QLD 4064 http://www.vrt.com.au ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] OpenNebula ceph snapshots problem.
System snapshot is only supported in KVM for qcow2 disks [1] so it won't work with your ceph datastore. If you want to use system snapshots and copy on write use the qcow2 datastore [2]. Cheers [1] http://docs.opennebula.org/4.4/user/virtual_resource_management/vm_guide_2.html#snapshotting [2] http://docs.opennebula.org/4.4/administration/storage/fs_ds.html#using-the-qcow2-transfer-driver On Wed, Apr 16, 2014 at 5:18 PM, Leszek Master keks...@gmail.com wrote: 1) I'm using OpenNebula 4.4 with ceph datastore. When i try to make snapshot i've got error: error: unsupported configuration: internal checkpoints require at least one disk to be selected for snapshot I'm using qcow2 images converted by opennebula to raw when uploaded. Can anyone help me ? 2) Also is there any way to schedule everyday snapshots of the vm ? 3) Is there any way to use copy-on-write in opennebula 4.4 ? ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org -- Javier Fontán Muiños Developer OpenNebula - The Open Source Toolkit for Data Center Virtualization www.OpenNebula.org | @OpenNebula | github.com/jfontan ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org