|
| 1 | +--- |
| 2 | +description: What is the recovery service feature and why you want it |
| 3 | +--- |
| 4 | +# Recovery service |
| 5 | + |
| 6 | +When you [create a new server](../howto/openstack/nova/new-server.md) in {{brand}} you will notice an option named **Recovery service**, which is enabled by default. |
| 7 | + |
| 8 | + |
| 9 | + |
| 10 | +Even if you choose to disable it for a particular server, keep in mind that you have the option to enable it at a later time. |
| 11 | + |
| 12 | + |
| 13 | + |
| 14 | +In the following, we explain what this option does, how it works in the background, and why you should consider enabling it. |
| 15 | + |
| 16 | +## What it is |
| 17 | + |
| 18 | +The *recovery service* feature is available via the {{gui}} and applies to servers and volumes that use our [Ceph](https://docs.ceph.com/) backend. |
| 19 | +That would be **all** servers but the ones of the `s` [flavor](../reference/flavors/index.md#compute-tiers). |
| 20 | + |
| 21 | +## How it works |
| 22 | + |
| 23 | +As soon as you enable the recovery service for a server or a single volume, you start getting snapshots for the corresponding [*RADOS Block Device* (RBD)](https://docs.ceph.com/en/latest/glossary/#term-Ceph-Block-Device) image. |
| 24 | + |
| 25 | +Those snapshots are created automatically once per day. |
| 26 | +By default, you have the snapshots of the last 10 days. |
| 27 | +Optionally, you may choose to keep snapshots for the past 30 days. |
| 28 | +The cost of a 30-day snapshot retention is 2× (**not** 3×) the cost of a 10-day snapshot retention. |
| 29 | + |
| 30 | +You can also make the snapshots immutable. |
| 31 | +In that case, you will not be able to delete snapshots during the retention period manually. |
| 32 | + |
| 33 | +Keep in mind that you cannot delete a volume with snapshots. |
| 34 | +So, as an example, if you have chosen a retention period of 30 days and also enabled immutability, then you will not be able to delete the volume before 30 days have passed. |
| 35 | + |
| 36 | +At any time, you may disable the recovery service, modify the retention period, or disable immutability for snapshots. |
| 37 | + |
| 38 | + |
| 39 | + |
| 40 | +Let's say, for instance, that you have enabled the recovery service and also the immutability feature for snapshots. |
| 41 | +At some point, you want to change the retention period or disable immutability altogether; |
| 42 | +you can do any of that. |
| 43 | +Later on, you decide you do not need the volume anymore; |
| 44 | +you disable the recovery service, wait for 10 or 30 days, and then delete the volume. |
| 45 | + |
| 46 | +## Why enable it |
| 47 | + |
| 48 | +Provided snapshots are available, you can restore a server or a single volume to any of those snapshots. |
| 49 | +For instance, you may discover that due to faulty application logic or simply a bug, you are now experiencing data corruption. |
| 50 | +Then, one of your options would be to [go back in time](../howto/openstack/nova/restore-srv-to-snap.md) by restoring one of the available snapshots and keep going from there. |
| 51 | + |
| 52 | +## Restoration time |
| 53 | + |
| 54 | +You should know that the recovery service feature creates *point-in-time* snapshots on the storage level. |
| 55 | +The time required to restore a server to a particular snapshot depends on its size. |
| 56 | +During restoration, the server is shut off. |
| 57 | +After the restore, you need to power the server back on manually. |
| 58 | +Although this whole process takes time analogous to volume size, as we pointed out, we should also note that it only takes seconds to complete on average. |
0 commit comments