Docker learning -Docker storage

Docker provides resources for the two kinds of data stored in a container:

• A mirror layer and the container layer storage driver management.
• Data Volume。

storage driver

Mirror docker layered structure as follows:

the concept of layers of Docker mirror layer is introduced, the production process of each step in image rub total Europe, will generate a new image layer

Container consists of the top container a writable layer, and a plurality of read-only image layer, the data stored in the container on these layers. The biggest characteristic of this hierarchy is the Copy-on-Write:

• The new data is placed directly on top of the container layer.
• Modifying existing data mirroring layer will start to copy data to the container layer, the modified data is stored directly in the container layer, the mirror layer remains unchanged.
• If you have multiple layers of the same file name, the user can only see the files that the top layer.
  The hierarchy to create a mirror and containers, sharing and distribution has become very efficient, which thanks to Docker storage driver. It is to achieve a unified view of the storage driver stack multiple layers of data and provide users with a single, consolidated thereafter.
  Docker supports a variety of storage driver, there AUFS, Device Mapper, Btrfs, OverlayFS , VFS and ZFS. They can achieve a layered architecture, at the same time and have their own characteristics.
  Docker will give priority to the use of Linux distributions default storage driver.
  Will select the default driver depending on the configuration of the current system when Docker installation. The default driver has the best stability, because the default driver through rigorous testing on the release.
  Run Docker infoto view the current system used by Storage driver
  

centos use the default driver overlay2, the underlying file system xfs, data stored in the respective layers / var / lib / docker

For some containers, such as busybox just a toolbox, you do not need to save the data for later use, directly to exit when finished, is stored in the container when the container is removed layers work together data also be deleted.

Docker Data Management

docker container persistent data stored in two general ways:
volume
the bind Mount

either volume or bind mount, a host file system directory or file which is essentially
either volume or bind mount, on which the stored data lifecycle after stand container, that container delete data on the volume or bind mount, is still there

Volume

Docker Host is a file system file or directory on Volume essence, the container can be directly mount the file system. Volume has the following characteristics:

• Volume is a directory or file, rather than not formatted disk (block device).
• The container may read and write data in the volume.
• volume data can be permanently saved, even if its containers have been destroyed.

Because the volume is actually part of docker host file system, so the volume of space depends on the capacity of the file system is not currently in use.

volume Caution:

• volume of content exist beyond the lifetime of the vessel; still there after deletion
• When mounted volume, mount not need to specify the source and a mount point can, Docker generates a directory for each volume at / var / lib / docker / volumes path, as a source mount
• If the container has a directory mount point to point, then the data in the directory is to copy the volume
• If the mount point to point container empty directory, it will automatically create the required directories.
• If mounted on a volume that does not exist in the container start, Dokcer automatically create volumes
• Volume reused when the container volume stamp authority can be set by the read-only parameter ro

bind mount

bind mount is host to mount on existing directories or files to the container.

bind mount实际上是一个inode替换的过程。

bind mount机制主要作用，允许一个目录或者文件（不是整个设备）挂载到一个指定的目录上，而且在该挂载点上进行任何的操作，只是发生在被挂载的目录或者文件上，而原挂载点的内容则会被隐藏起来不受影响。

bind mount使用注意事项：

• 容器运行过程中，对bind mount目录中改动的数据，将被保存，删除容器后，bind mount中的数据任然存在。
• bind mount可以挂载在一个目录到容器，也可以挂载一个文件到容器，但必须要指定的目录或文件的路径，即mount源,当然也必须指定mount point，这也限制了容器的可移植性
• 若将bind mount也绑定到容器上的某非空目录下，则会隐藏容器目录下的现有内容，若不希望容器的整个目录被覆盖，可单独挂载某个文件
• 若mount源指向的文件或者目录在宿主机上不存在，则会自动创建
• bind mount时，可以通过ro参数将容器对数据的权限设置为只读，设置ro参数后，容器无法对数据进行修改，但宿主机依旧有权修改其内容。

bind mount的用法是使用-v选项将host已经存在的目录或者文件mount到容器
如下所示：

-v 的格式为 <host path>:<container path>。/usr/local/apache2/htdocs 就是 apache server 存放静态文件的地方。由于 /usr/local/apache2/htdocs 已经存在，原有数据会被隐藏起来，取而代之的是 host $HOME/htdocs/ 中的数据，这与 linux mount 命令的行为是一致的。

数据共享

数据共享是volume的关键特性，主机与容器数据共享：

• bind mount:将host上的目录或者文件mount到容器中
• volume:将Host上的数据copy到容器的volume中
  容器间的数据共享：
  bind mount:将host上目录或文件mount到多个容器中
  volume:将volume挂载到多个容器
  volume container:先通过volume或bind mount将数据挂载到一个container中，其他容器再引用这个container中的数据

volume container是专门为其他容器提供volume的容器。

volume生命周期管理

备份

因为 volume 实际上是 host 文件系统中的目录和文件，所以 volume 的备份实际上是对文件系统的备份

恢复

volume 的恢复也很简单，如果数据损坏了，直接用之前备份的数据拷贝

迁移

如果使用更新版本的 Registry，这就涉及到数据迁移，方法是：

• docker stop 当前 Registry 容器。
• 启动新版本容器并 mount 原有 volume。

销毁

volume删除后数据是找不回来的，注意
docker 不会销毁 bind mount，删除数据的工作只能由 host 负责,再在执行 docker rm 删除容器时可以带上 -v 参数，docker 会将容器使用到的 volume 一并删除，但前提是没有其他容器 mount 该 volume。

操作实验

volume挂载操作

Create a volume, and mounts a container httpd
docker run -d -p 8080:80 -v /usr/local/apache2/htdocs httpd
-v to mount it to the container httpd

-v format <host path>: <container path >. / usr / local / apache2 / htdocs is where the Apache Server to store static files,
due to the / usr / local / apache2 / htdocs already exists, the original data will be hidden and replaced by the data host $ HOME / htdocs / in.
Check the volume of information

docker volume ls

See mount the container volume information, and to obtain volume path, Type = volume

docker inspect d5db6a048612

View volume of data

 cd /var/lib/docker/volumes/6189c90831d019229a2e8593453fe1c334faec1fcc56db80b9f99773d21c9c55/_data

See the corresponding data in the container, the results: data = data container volume

Index.html file contents into the container update

docker exec -it d5db6a048612 bash
cd htdocs && echo "update the index" > index.html

View the contents of the volume again, it has been synchronized with the new

result data sharing and Host of container.

Force the removal of the container, and then view the data in the volume, you can see still there

docker rm -f d5db6a048612

bind mount

The host computer / root / htdocs directory in a read-only mount the container to a named httpd httpd1 mapped port 8081

docker run --name httpd1 -d -p 8081:80 -v /root/htdocs:/usr/local/apache2/htdocs:ro httpd

Mount View container information. Type = bind

docker inspect httpd1

Index.html file with new data on host host, confirms the data httpd1 container also together with the new

result, container and Host realizes data sharing

Httpd1 into the container update index.html file data, suggesting Read-only

The host / root / htdocs mounted to the container named http httpd2 mapped port 8082 is not provided ro

docker run --name httpd2 -d -p 8082:80 -v /root/htdocs/:/usr/local/apache2/htdocs httpd

Httpd2 into the container update index.html

docker exec -it httpd2 bash

See each host, HTTP1, the data http2, three consistent with the data

at this time confirms shared data between the containers