Re: [openstack-dev] [cinder]Do we have plan for restoring on-line volume?
Yes, John, It's for boot from volume case. We can't restore the instance root disk now with backups, although we can make backup with in-use volumes. Snapshot is same, we can take a snapshot for in-use volumes, but we only can create a new volume from it. You know in some cases users don't want to change there instance's id, ip address, or mac etc, even with DR scenarios, when fail back, users may still want their old instances with old ip address and other stuff So I feel maybe a better way is that nova can support detach boot volume. 2015-08-12 4:46 GMT+08:00 John Griffith john.griffi...@gmail.com: On Mon, Aug 10, 2015 at 9:54 PM, hao wang sxmatch1...@gmail.com wrote: @Duncan and John, I have a idea about this issue, if volume is in-use but the instance which it's attached to is power-off, so can we avoid data corruption? If it's true, cinder could free the available restriction, let user decide if they want to restore the boot disk, and they should shut down the instance first and then restore volume. 2015-08-11 11:30 GMT+08:00 hao wang sxmatch1...@gmail.com: @Duncan and John, thank you both and I have understood this. The reason I ask this question is we can't restore the boot volume of instance which is boot-from-volume, since Nova can't detach the boot disk now. So we need to figure out a way to solve this issue. Maybe we should let nova can detach the boot disk, but there are also some problems, like what state should be set after detach the boot disk? Can user reboot/stop/migrate the instance after detach the boot disk? etc. Anyway, I feel it may be a long way to solve this issue properly. 2015-08-11 0:13 GMT+08:00 John Griffith john.griffi...@gmail.com: On Mon, Aug 10, 2015 at 3:47 AM, hao wang sxmatch1...@gmail.com wrote: Sorry if I missed something, since now we have supported backup in-use volumes in L, so I wonder is there some plan or design to support restoring on-line volumes. Thanks. -- Best Wishes For You! __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev No, a restore operation is going to require that the volume be taken offline for a number of reasons. Even with multi-attach if/when it ever becomes a reality you can't write to the volume while it's in use by another Instance (corruption, file system updates etc etc) unless you're using a shared FS or clustered File System. That's a use case that I don't think has a ton of value and it's not compelling enough to introduce all of the issues that come along with it IMO. Thanks, John __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Best Wishes For You! -- Best Wishes For You! __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev Wouldn't a detach be less impact than powering off the instance? I don't understand why this is desired? Is there a compelling reason perhaps that I'm not aware of for this? Oh... maybe you are thinking specifically for the boot from volume case here? I guess if that's the case I could see why you may desire to keep the instance ID etc after a restore. Although I wish we could all just agree that things in Clouds are dynamic and things like their ID can change and just deal with it. For point in time recovery type things I still think snapshots are great, for full DR, backups are preferred and with full DR scenarios I don't think it's unreasonable to require a little extra effort. __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Best Wishes For You! __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [cinder]Do we have plan for restoring on-line volume?
On Mon, Aug 10, 2015 at 3:47 AM, hao wang sxmatch1...@gmail.com wrote: Sorry if I missed something, since now we have supported backup in-use volumes in L, so I wonder is there some plan or design to support restoring on-line volumes. Thanks. -- Best Wishes For You! __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev No, a restore operation is going to require that the volume be taken offline for a number of reasons. Even with multi-attach if/when it ever becomes a reality you can't write to the volume while it's in use by another Instance (corruption, file system updates etc etc) unless you're using a shared FS or clustered File System. That's a use case that I don't think has a ton of value and it's not compelling enough to introduce all of the issues that come along with it IMO. Thanks, John __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [cinder]Do we have plan for restoring on-line volume?
@Duncan and John, thank you both and I have understood this. The reason I ask this question is we can't restore the boot volume of instance which is boot-from-volume, since Nova can't detach the boot disk now. So we need to figure out a way to solve this issue. Maybe we should let nova can detach the boot disk, but there are also some problems, like what state should be set after detach the boot disk? Can user reboot/stop/migrate the instance after detach the boot disk? etc. Anyway, I feel it may be a long way to solve this issue properly. 2015-08-11 0:13 GMT+08:00 John Griffith john.griffi...@gmail.com: On Mon, Aug 10, 2015 at 3:47 AM, hao wang sxmatch1...@gmail.com wrote: Sorry if I missed something, since now we have supported backup in-use volumes in L, so I wonder is there some plan or design to support restoring on-line volumes. Thanks. -- Best Wishes For You! __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev No, a restore operation is going to require that the volume be taken offline for a number of reasons. Even with multi-attach if/when it ever becomes a reality you can't write to the volume while it's in use by another Instance (corruption, file system updates etc etc) unless you're using a shared FS or clustered File System. That's a use case that I don't think has a ton of value and it's not compelling enough to introduce all of the issues that come along with it IMO. Thanks, John __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Best Wishes For You! __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [cinder]Do we have plan for restoring on-line volume?
@Duncan and John, I have a idea about this issue, if volume is in-use but the instance which it's attached to is power-off, so can we avoid data corruption? If it's true, cinder could free the available restriction, let user decide if they want to restore the boot disk, and they should shut down the instance first and then restore volume. 2015-08-11 11:30 GMT+08:00 hao wang sxmatch1...@gmail.com: @Duncan and John, thank you both and I have understood this. The reason I ask this question is we can't restore the boot volume of instance which is boot-from-volume, since Nova can't detach the boot disk now. So we need to figure out a way to solve this issue. Maybe we should let nova can detach the boot disk, but there are also some problems, like what state should be set after detach the boot disk? Can user reboot/stop/migrate the instance after detach the boot disk? etc. Anyway, I feel it may be a long way to solve this issue properly. 2015-08-11 0:13 GMT+08:00 John Griffith john.griffi...@gmail.com: On Mon, Aug 10, 2015 at 3:47 AM, hao wang sxmatch1...@gmail.com wrote: Sorry if I missed something, since now we have supported backup in-use volumes in L, so I wonder is there some plan or design to support restoring on-line volumes. Thanks. -- Best Wishes For You! __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev No, a restore operation is going to require that the volume be taken offline for a number of reasons. Even with multi-attach if/when it ever becomes a reality you can't write to the volume while it's in use by another Instance (corruption, file system updates etc etc) unless you're using a shared FS or clustered File System. That's a use case that I don't think has a ton of value and it's not compelling enough to introduce all of the issues that come along with it IMO. Thanks, John __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Best Wishes For You! -- Best Wishes For You! __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [cinder]Do we have plan for restoring on-line volume?
Does restoring a volume online make any logical sense? Generally operating systems don't like it when a mounted volume changes contents, and a restore, being a slow sequential write is likely to be even worse. Since you have to unmount anyway, attaching a new volume should not be an issue. Just because a feature is technically possible doesn't mean it is a good idea, particularly when the use of that feature under normal circumstances causes data corruption. We have plenty of code paths in cinder already that are hard to exercise fully and buggy (e.g migration) without adding more. On 10 Aug 2015 10:49, hao wang sxmatch1...@gmail.com wrote: Sorry if I missed something, since now we have supported backup in-use volumes in L, so I wonder is there some plan or design to support restoring on-line volumes. Thanks. -- Best Wishes For You! __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
