8c9e5e5e05
The way thin-pool right now is designed, user space is supposed to keep track of what device ids have already been used. If user space tries to create a new thin/snap device and device id has already been used, thin pool retuns -EEXIST. Upon receiving -EEXIST, current docker implementation simply tries the NextDeviceId++ and keeps on doing this till it finds a free device id. This approach has two issues. - It is little suboptimal. - If device id already exists, current kenrel implementation spits out a messsage on console. [17991.140135] device-mapper: thin: Creation of new snapshot 33 of device 3 failed. Here kenrel is trying to tell user that device id 33 has already been used. And this shows up for every device id docker tries till it reaches a point where device ids are not used. So if there are thousands of container and one is trying to create a new container after fresh docker start, expect thousands of such warnings to flood console. This patch saves the NextDeviceId in a file in /var/lib/docker/devmapper/metadata/deviceset-metadata and reads it back when docker starts. This way we don't retry lots of device ids which have already been used. There might be some device ids which are free but we will get back to them once device numbers wrap around (24bit limit on device ids). This patch should cut down on number of kernel warnings. Notice that I am creating a deviceset metadata file which is a global file for this pool. So down the line if we need to save more data we should be able to do that. Signed-off-by: Vivek Goyal <vgoyal@redhat.com> |
||
|---|---|---|
| .. | ||
| MAINTAINERS | ||
| README.md | ||
| attach_loopback.go | ||
| deviceset.go | ||
| devmapper.go | ||
| devmapper_doc.go | ||
| devmapper_log.go | ||
| devmapper_test.go | ||
| devmapper_wrapper.go | ||
| driver.go | ||
| ioctl.go | ||
| mount.go | ||
README.md
devicemapper - a storage backend based on Device Mapper
Theory of operation
The device mapper graphdriver uses the device mapper thin provisioning
module (dm-thinp) to implement CoW snapshots. For each devicemapper
graph location (typically /var/lib/docker/devicemapper, $graph below)
a thin pool is created based on two block devices, one for data and
one for metadata. By default these block devices are created
automatically by using loopback mounts of automatically created sparse
files.
The default loopback files used are $graph/devicemapper/data and
$graph/devicemapper/metadata. Additional metadata required to map
from docker entities to the corresponding devicemapper volumes is
stored in the $graph/devicemapper/json file (encoded as Json).
In order to support multiple devicemapper graphs on a system, the thin
pool will be named something like: docker-0:33-19478248-pool, where
the 0:33 part is the minor/major device nr and 19478248 is the
inode number of the $graph directory.
On the thin pool, docker automatically creates a base thin device,
called something like docker-0:33-19478248-base of a fixed
size. This is automatically formatted with an empty filesystem on
creation. This device is the base of all docker images and
containers. All base images are snapshots of this device and those
images are then in turn used as snapshots for other images and
eventually containers.
options
The devicemapper backend supports some options that you can specify
when starting the docker daemon using the --storage-opt flags.
This uses the dm prefix and would be used something like docker -d --storage-opt dm.foo=bar.
Here is the list of supported options:
-
dm.basesizeSpecifies the size to use when creating the base device, which limits the size of images and containers. The default value is 10G. Note, thin devices are inherently "sparse", so a 10G device which is mostly empty doesn't use 10 GB of space on the pool. However, the filesystem will use more space for the empty case the larger the device is. Warning: This value affects the system-wide "base" empty filesystem that may already be initialized and inherited by pulled images. Typically, a change to this value will require additional steps to take effect: 1) stop
docker -d, 2)rm -rf /var/lib/docker, 3) startdocker -d.Example use:
docker -d --storage-opt dm.basesize=20G -
dm.loopdatasizeSpecifies the size to use when creating the loopback file for the "data" device which is used for the thin pool. The default size is 100G. Note that the file is sparse, so it will not initially take up this much space.
Example use:
docker -d --storage-opt dm.loopdatasize=200G -
dm.loopmetadatasizeSpecifies the size to use when creating the loopback file for the "metadadata" device which is used for the thin pool. The default size is 2G. Note that the file is sparse, so it will not initially take up this much space.
Example use:
docker -d --storage-opt dm.loopmetadatasize=4G -
dm.fsSpecifies the filesystem type to use for the base device. The supported options are "ext4" and "xfs". The default is "ext4"
Example use:
docker -d --storage-opt dm.fs=xfs -
dm.mkfsargSpecifies extra mkfs arguments to be used when creating the base device.
Example use:
docker -d --storage-opt "dm.mkfsarg=-O ^has_journal" -
dm.mountoptSpecifies extra mount options used when mounting the thin devices.
Example use:
docker -d --storage-opt dm.mountopt=nodiscard -
dm.datadevSpecifies a custom blockdevice to use for data for the thin pool.
If using a block device for device mapper storage, ideally both datadev and metadatadev should be specified to completely avoid using the loopback device.
Example use:
docker -d --storage-opt dm.datadev=/dev/sdb1 --storage-opt dm.metadatadev=/dev/sdc1 -
dm.metadatadevSpecifies a custom blockdevice to use for metadata for the thin pool.
For best performance the metadata should be on a different spindle than the data, or even better on an SSD.
If setting up a new metadata pool it is required to be valid. This can be achieved by zeroing the first 4k to indicate empty metadata, like this:
``dd if=/dev/zero of=$metadata_dev bs=4096 count=1```
Example use:
docker -d --storage-opt dm.datadev=/dev/sdb1 --storage-opt dm.metadatadev=/dev/sdc1 -
dm.blocksizeSpecifies a custom blocksize to use for the thin pool. The default blocksize is 64K.
Example use:
docker -d --storage-opt dm.blocksize=512K -
dm.blkdiscardEnables or disables the use of blkdiscard when removing devicemapper devices. This is enabled by default (only) if using loopback devices and is required to res-parsify the loopback file on image/container removal.
Disabling this on loopback can lead to much faster container removal times, but will make the space used in /var/lib/docker directory not be returned to the system for other use when containers are removed.
Example use:
docker -d --storage-opt dm.blkdiscard=false