Actions
Task #7649
closedSketch a VM backup & restore
Start date:
01/27/2020
Due date:
% Done:
100%
Estimated time:
(Total: 0.00 h)
PM Check date:
Description
- User wants to have possibility to roll back in time
Implementation¶
- We can use ceph snapshots
- We might need/want to signal qemu before taking the snapshot
- how?
- Difference between uncloud and opennebula?
User facing API / CLI¶
- User can set a maximum amount of snapshots (?)
- Useful for auto rotate
- User can configure a snapshot
- User can list all snapshots
- User can delete snapshots
Ideas for an CLI¶
ungleich-cli --username xx --password yy list-vms ungleich-cli --username xx --password yy vm-snaphot-create <vmid> ungleich-cli --username xx --password yy vm-snaphot-list <vmid> ungleich-cli --username xx --password yy vm-snaphot-restore <vmid> <snapshot> ungleich-cli --username xx --password yy vm-snaphot-delete <vmid> <snapshot>
Billing¶
Regular pricing - it's the same usage
- SSD storage 10GB/month@3.5CHF
- HDD storage 100GB/month@1.5CHF
Billed per gigabyte per second (?) basis (?)
Per second calculation: SSD¶
>>> (3.5/10.)/30/24/3600
1.3503086419753083e-07
>>> per_gb_per_second=((3.5/10.)/30/24/3600)
>>> 1.36*10**(-7) * 10 * 86400 * 30
3.5251200000000003
>>> "{:0.9f}".format(1.36*10**(-7))
'0.000000136'
>>> 1.4*10**(-7) * 10 * 86400 * 30
3.628799999999999
- Could offer it for 0.00000014 CHF per GB per second
- Monthly would stay the same
- User wants to have possibility to roll back in time
Implementation¶
- We can use ceph snapshots
- We might need/want to signal qemu before taking the snapshot
- how?
- Difference between uncloud and opennebula?
User facing API / CLI¶
- User can set a maximum amount of snapshots (?)
- Useful for auto rotate
- User can configure a snapshot
- User can list all snapshots
- User can delete snapshots
Billing¶
Regular pricing - it's the same usage
- SSD storage 10GB/month@3.5CHF
- HDD storage 100GB/month@1.5CHF
Billed per gigabyte per second (?) basis (?)
Per second calculation: SSD¶
>>> (3.5/10.)/30/24/3600
1.3503086419753083e-07
>>> per_gb_per_second=((3.5/10.)/30/24/3600)
>>> 1.36*10**(-7) * 10 * 86400 * 30
3.5251200000000003
>>> "{:0.9f}".format(1.36*10**(-7))
'0.000000136'
>>> 1.4*10**(-7) * 10 * 86400 * 30
3.628799999999999
- Could offer it for 0.00000014 CHF per GB per second
- Monthly would stay the same
Per second calculation: HDD¶
>>> per_gb_per_second=((1.5/100.)/30/24/3600)
>>> per_gb_per_second
5.787037037037037e-09
>>>
>>> "{:0.10f}".format(per_gb_per_second)
'0.0000000058'
>>> 0.0000000058* 100 * 86400 * 30
1.5033599999999998
- Could offer for 0.0000000058 per gb per second
- However units will be smaller than 100gb packages
Testing¶
- Using my private VM
- Locating the disk image
[10:43:59] red2.place5:~# rbd -p one ls | grep 10761 one-60-10761-0 [10:44:14] red2.place5:~#
- Does not show all attached disks -> need to go via opennebula api
Per second calculation: HDD¶
>>> per_gb_per_second=((1.5/100.)/30/24/3600)
>>> per_gb_per_second
5.787037037037037e-09
>>>
>>> "{:0.10f}".format(per_gb_per_second)
'0.0000000058'
Testing¶
- Using my private VM
- Locating the disk image
[10:43:59] red2.place5:~# rbd -p one ls | grep 10761 one-60-10761-0 [10:44:14] red2.place5:~#
- Does not show all attached disks -> need to go via opennebula api
Steps¶
- Create a PoC for some VMs
- record creation and deletion time is needed for billing later
- Implement a CLI to be used by staff (i.e. no customer auth)
- Integrate it into billing
Updated by Nico Schottelius almost 5 years ago
- Description updated (diff)
Ceph¶
It seems ceph even has a limit support:
snap create (snap add) Create a snapshot.
snap limit clear Remove snapshot limit.
snap limit set Limit the number of snapshots.
snap list (snap ls) Dump list of image snapshots.
snap protect Prevent a snapshot from being deleted.
snap purge Delete all snapshots.
snap remove (snap rm) Delete a snapshot.
snap rename Rename a snapshot.
snap rollback (snap revert) Rollback image to snapshot.
snap unprotect Allow a snapshot to be deleted.
[10:48:18] red2.place5:~# rbd snap create one/one-60-10761-0@20200124_1048
[10:49:07] red2.place5:~# rbd snap ls one/one-60-10761-0
SNAPID NAME SIZE TIMESTAMP
124 20200124_1048 500GiB Fri Jan 24 10:49:07 2020
[10:49:14] red2.place5:~#
Updated by Nico Schottelius almost 5 years ago
Snapshotting:
[10:52:32] red2.place5:~# rbd rm one/nicotest 2020-01-24 11:14:57.886218 7fc527460700 -1 librbd::image::RemoveRequest: 0x56525faae5f0 check_image_snaps: image has snapshots - not removing Removing image: 0% complete...failed. rbd: image has snapshots - these must be deleted with 'rbd snap purge' before the image can be removed. [11:14:57] red2.place5:~# rbd snap purge one/nicotest Removing all snapshots: 100% complete...done. [11:15:41] red2.place5:~# rbd rm one/nicotest Removing image: 100% complete...done. [11:15:43] red2.place5:~#
Updated by Nico Schottelius almost 5 years ago
OpenNebula testing¶
[11:17:33] server1.place5:~# onevm list | grep nico-vm 10761 nico.schottelius@ungleich.ch users nico-vm.schotte runn 5.05 2.1G server1.place5.ungleich.ch 918d 16h56 [11:18:07] server1.place5:~# onevm snapshot-create 10761 [11:18:17] server1.place5:~#
- Did not create anything
- Using the webui creating a snapshot per disk works
Updated by Ahmed Bilal almost 5 years ago
- Assignee changed from Nico Schottelius to Ahmed Bilal
Updated by Ahmed Bilal almost 5 years ago
The functionality part is complete beside `ungleich-cli` and can be found at https://code.ungleich.ch/ahmedbilal/vm-cli-services. I am ready to give a test-run tour whenever you want.
Updated by Nico Schottelius 11 months ago
- Status changed from In Progress to Rejected
Actions