Use snapshot to implement HDFs file backup and recovery combat

Enable backup of files on HDFs via snapshot

API address please see http://archive.cloudera.com/cdh5/cdh/5/hadoop-2.5.0-cdh5.2.0/hadoop-project-dist/hadoop-hdfs/HdfsSnapshots.html

==========================================================================================

1. Allow snapshot creation

First, execute the command below the folder where you want to make the backup, allowing the folder to create a snapshot

HDFs Dfsadmin-allowsnapshot <path>

Example: HDFs Dfsadmin-allowsnapshot/workspace/linlin

This command appears to prove that the snapshot was allowed to be created successfully

=================================================================================================

2. Create a Snapshot

The next step is to start a backup of this folder

HDFs dfs-createsnapshot <path> [name]

such as HDFs Dfs-createsnapshot/workspace/linlin BAK1

This command appears to prove that the snapshot has been created successfully

At this point we can consider whether you can create a snapshot in the Linlin subdirectory

HDFs ' on the directory structure

Then try to create a snapshot on the Snaptest

HDFs dfs-createsnapshot/workspace/linlin/snaptest BAK2

Error, visible, can only be in the directory under your permission to create a snapshot;

The first snapshot bak1 when there is no Snaptest folder, now more Snaptest folder, and then create a snapshot

If you still use the

HDFs Dfs-createsnapshot/workspace/linlin BAK1

There is an error indicating that the snapshot name already exists

Performing HDFs Dfs-createsnapshot/workspace/linlin BAK2

Creation success;

==============================================================================================================

3. View Snapshots

See all Snapshottable

   HDFs Lssnapshottabledir

To view all directories that were once allowed to create a snapshot

View files under the current snapshot  when Hadoop sanpshot creates a snapshot, the default folder is. Snapshot you must add the. Snapshot to see what's back there;

. Snapshot is a product of later Hadoop, so it is not possible to create a snapshot if a folder is named snapshot keyword;

Execute command HDFs dfs-ls/workspace/linlin/.snapshot/

Can see three backups under this snapshot are Bak1, BAK2, Linlin, respectively.

===========================================================================================================

4. Compare Snapshots

Compare snapshots to see the difference between backup files between two snapshots

Execute command

HDFs Snapshotdiff <path> <fromSnapshot> <toSnapshot>

Execute command HDFs Snapshotdiff  

Results Results:

+	The file/directory has been created.
-	The file/directory has been deleted.
M	The file/directory has been modified.
R	The file/directory has been renamed.

The presence of M here means I have modified the Linlin folder, and the + represents a new folder Snaptest

=============================================================================================================== =============

5. Recovering snapshots

To recover a snapshot:

hdfs dfs -cp <path> <path>

Example: HDFs dfs-cp/workspace/linlin/.snapshot/bak2/snaptest/workspace

To view the HDFs directory:

has been successfully restored to Workspace

=============================================================================================================== ===================

Off Topic:

We can try to delete the folder where the snapshot was created: it cannot be deleted, it will prompt

A snapshot cannot be deleted, proving that a folder cannot be deleted or moved if a snapshot is created under a folder

Original: HTTP://WWW.NOSQLCN.COM/SHOWARTICLE/23

Use snapshot to implement HDFs file backup and recovery combat