docs/scheduler/filter

page_title	page_description	page_keywords
Docker Swarm filters	Swarm filters	docker, swarm, clustering, filters

Filters

The Docker Swarm scheduler comes with multiple filters.

The following filters are currently used to schedule containers on a subset of nodes:

• Constraint
• Affinity
• Port
• Dependency
• Health

You can choose the filter(s) you want to use with the --filter flag of swarm manage

Constraint Filter

Constraints are key/value pairs associated to particular nodes. You can see them as node tags.

When creating a container, the user can select a subset of nodes that should be considered for scheduling by specifying one or more sets of matching key/value pairs.

This approach has several practical use cases such as:

• Selecting specific host properties (such as storage=ssd, in order to schedule containers on specific hardware).
• Tagging nodes based on their physical location (region=us-east, to force containers to run on a given location).
• Logical cluster partitioning (environment=production, to split a cluster into sub-clusters with different properties).

To tag a node with a specific set of key/value pairs, one must pass a list of --label options at docker startup time.

For instance, let's start node-1 with the storage=ssd label:

$ docker -d --label storage=ssd
$ swarm join --addr=192.168.0.42:2375 token://XXXXXXXXXXXXXXXXXX

Again, but this time node-2 with storage=disk:

$ docker -d --label storage=disk
$ swarm join --addr=192.168.0.43:2375 token://XXXXXXXXXXXXXXXXXX

Once the nodes are registered with the cluster, the master pulls their respective tags and will take them into account when scheduling new containers.

Let's start a MySQL server and make sure it gets good I/O performance by selecting nodes with flash drives:

$ docker run -d -P -e constraint:storage==ssd --name db mysql
f8b693db9cd6

$ docker ps
CONTAINER ID        IMAGE               COMMAND             CREATED                  STATUS              PORTS                           NODE        NAMES
f8b693db9cd6        mysql:latest        "mysqld"            Less than a second ago   running             192.168.0.42:49178->3306/tcp    node-1      db

In this case, the master selected all nodes that met the storage=ssd constraint and applied resource management on top of them, as discussed earlier. node-1 was selected in this example since it's the only host running flash.

Now we want to run an Nginx frontend in our cluster. However, we don't want flash drives since we'll mostly write logs to disk.

$ docker run -d -P -e constraint:storage==disk --name frontend nginx
963841b138d8

$ docker ps
CONTAINER ID        IMAGE               COMMAND             CREATED                  STATUS              PORTS                           NODE        NAMES
963841b138d8        nginx:latest        "nginx"             Less than a second ago   running             192.168.0.43:49177->80/tcp      node-2      frontend
f8b693db9cd6        mysql:latest        "mysqld"            Up About a minute        running             192.168.0.42:49178->3306/tcp    node-1      db

The scheduler selected node-2 since it was started with the storage=disk label.

Standard Constraints

Additionally, a standard set of constraints can be used when scheduling containers without specifying them when starting the node. Those tags are sourced from docker info and currently include:

• node ID or node Name (using key "node")
• storagedriver
• executiondriver
• kernelversion
• operatingsystem

Affinity Filter

Containers

You can schedule 2 containers and make the container #2 next to the container #1.

$ docker run -d -p 80:80 --name front nginx
 87c4376856a8

$ docker ps
CONTAINER ID        IMAGE               COMMAND             CREATED                  STATUS              PORTS                           NODE        NAMES
87c4376856a8        nginx:latest        "nginx"             Less than a second ago   running             192.168.0.42:80->80/tcp         node-1      front

Using -e affinity:container==front will schedule a container next to the container front. You can also use IDs instead of name: -e affinity:container==87c4376856a8

$ docker run -d --name logger -e affinity:container==front logger
 87c4376856a8

$ docker ps
CONTAINER ID        IMAGE               COMMAND             CREATED                  STATUS              PORTS                           NODE        NAMES
87c4376856a8        nginx:latest        "nginx"             Less than a second ago   running             192.168.0.42:80->80/tcp         node-1      front
963841b138d8        logger:latest       "logger"            Less than a second ago   running                                             node-1      logger

The logger container ends up on node-1 because its affinity with the container front.

Images

You can schedule a container only on nodes where a specific image is already pulled.

$ docker -H node-1:2375 pull redis
$ docker -H node-2:2375 pull mysql
$ docker -H node-3:2375 pull redis

Here only node-1 and node-3 have the redis image. Using -e affinity:image=redis we can schedule container only on these 2 nodes. You can also use the image ID instead of its name.

$ docker run -d --name redis1 -e affinity:image==redis redis
$ docker run -d --name redis2 -e affinity:image==redis redis
$ docker run -d --name redis3 -e affinity:image==redis redis
$ docker run -d --name redis4 -e affinity:image==redis redis
$ docker run -d --name redis5 -e affinity:image==redis redis
$ docker run -d --name redis6 -e affinity:image==redis redis
$ docker run -d --name redis7 -e affinity:image==redis redis
$ docker run -d --name redis8 -e affinity:image==redis redis

$ docker ps
CONTAINER ID        IMAGE               COMMAND             CREATED                  STATUS              PORTS                           NODE        NAMES
87c4376856a8        redis:latest        "redis"             Less than a second ago   running                                             node-1      redis1
1212386856a8        redis:latest        "redis"             Less than a second ago   running                                             node-1      redis2
87c4376639a8        redis:latest        "redis"             Less than a second ago   running                                             node-3      redis3
1234376856a8        redis:latest        "redis"             Less than a second ago   running                                             node-1      redis4
86c2136253a8        redis:latest        "redis"             Less than a second ago   running                                             node-3      redis5
87c3236856a8        redis:latest        "redis"             Less than a second ago   running                                             node-3      redis6
87c4376856a8        redis:latest        "redis"             Less than a second ago   running                                             node-3      redis7
963841b138d8        redis:latest        "redis"             Less than a second ago   running                                             node-1      redis8

As you can see here, the containers were only scheduled on nodes with the redis image already pulled.

Expression Syntax

An affinity or a constraint expression consists of a key and a value. A key must conform the alpha-numeric pattern, with the leading alphabet or underscore.

A value must be one of the following:

• An alpha-numeric string, dots, hyphens, and underscores.
• A globbing pattern, i.e., abc*.
• A regular expression in the form of /regexp/. We support the Go's regular expression syntax.

Currently Swarm supports the following affinity/constraint operators: == and !=.

For example,

• constraint:node==node1 will match node node1.
• constraint:node!=node1 will match all nodes, except node1.
• constraint:region!=us* will match all nodes outside the regions prefixed with us.
• constraint:node==/node[12]/ will match nodes node1 and node2.
• constraint:node==/node\d/ will match all nodes with node + 1 digit.
• constraint:node!=/node-[01]/ will match all nodes, except node-0 and node-1.
• constraint:node!=/foo\[bar\]/ will match all nodes, except foo[bar]. You can see the use of escape characters here.
• constraint:node==/(?i)node1/ will match node node1 case-insensitive. So NoDe1 or NODE1 will also match.

Soft Affinities/Constraints

By default, affinities and constraints are hard enforced. If an affinity or constraint is not met, the container won't be scheduled. With soft affinities/constraints the scheduler will try to meet the rule. If it is not met, the scheduler will discard the filter and schedule the container according to the scheduler's strategy.

Soft affinities/constraints are expressed with a ~ in the expression, for example:

$ docker run -d --name redis1 -e affinity:image==~redis redis

If none of the nodes in the cluster has the image redis, the scheduler will discard the affinity and schedule according to the strategy.

$ docker run -d --name redis2 -e constraint:region==~us* redis

If none of the nodes in the cluster belongs to the us region, the scheduler will discard the constraint and schedule according to the strategy.

$ docker run -d --name redis5 -e affinity:container!=~redis* redis

The affinity filter will be used to schedule a new redis5 container to a different node that doesn't have a container with the name that satisfies redis*. If each node in the cluster has a redis* container, the scheduler will discard the affinity rule and schedule according to the strategy.

Port Filter

With this filter, ports are considered unique resources.

$ docker run -d -p 80:80 nginx
87c4376856a8

$ docker ps
CONTAINER ID    IMAGE               COMMAND         PORTS                       NODE        NAMES
87c4376856a8    nginx:latest        "nginx"         192.168.0.42:80->80/tcp     node-1      prickly_engelbart

Docker cluster selects a node where the public 80 port is available and schedules a container on it, in this case node-1.

Attempting to run another container with the public 80 port will result in the cluster selecting a different node, since that port is already occupied on node-1:

$ docker run -d -p 80:80 nginx
963841b138d8

$ docker ps
CONTAINER ID        IMAGE          COMMAND        PORTS                           NODE        NAMES
963841b138d8        nginx:latest   "nginx"        192.168.0.43:80->80/tcp         node-2      dreamy_turing
87c4376856a8        nginx:latest   "nginx"        192.168.0.42:80->80/tcp         node-1      prickly_engelbart

Again, repeating the same command will result in the selection of node-3, since port 80 is neither available on node-1 nor node-2:

$ docker run -d -p 80:80 nginx
963841b138d8

$ docker ps
CONTAINER ID   IMAGE               COMMAND        PORTS                           NODE        NAMES
f8b693db9cd6   nginx:latest        "nginx"        192.168.0.44:80->80/tcp         node-3      stoic_albattani
963841b138d8   nginx:latest        "nginx"        192.168.0.43:80->80/tcp         node-2      dreamy_turing
87c4376856a8   nginx:latest        "nginx"        192.168.0.42:80->80/tcp         node-1      prickly_engelbart

Finally, Docker Swarm will refuse to run another container that requires port 80 since not a single node in the cluster has it available:

$ docker run -d -p 80:80 nginx
2014/10/29 00:33:20 Error response from daemon: no resources available to schedule container

Port filter in Host Mode

Docker in the host mode, running with --net=host, differs from the default bridge mode as the host mode does not perform any port binding. So, it require that you explicitly expose one or more port numbers (using EXPOSE in the Dockerfile or --expose on the command line). Swarm makes use of this information in conjunction with the host mode to choose an available node for a new container.

For example, the following commands start nginx on 3-node cluster.

$ docker run -d --expose=80 --net=host nginx
640297cb29a7
$ docker run -d --expose=80 --net=host nginx
7ecf562b1b3f
$ docker run -d --expose=80 --net=host nginx
09a92f582bc2

Port binding information will not be available through the docker ps command because all the nodes are started in the host mode.

$ docker ps
CONTAINER ID        IMAGE               COMMAND                CREATED                  STATUS              PORTS               NAMES
640297cb29a7        nginx:1             "nginx -g 'daemon of   Less than a second ago   Up 30 seconds                           box3/furious_heisenberg
7ecf562b1b3f        nginx:1             "nginx -g 'daemon of   Less than a second ago   Up 28 seconds                           box2/ecstatic_meitner
09a92f582bc2        nginx:1             "nginx -g 'daemon of   46 seconds ago           Up 27 seconds                           box1/mad_goldstine

The swarm will refuse the operation when trying to instantiate the 4th container.

$  docker run -d --expose=80 --net=host nginx
FATA[0000] Error response from daemon: unable to find a node with port 80/tcp available in the Host mode

However port binding to the different value, e.g. 81, is still allowed.

$  docker run -d -p 81:80 nginx:latest
832f42819adc
$  docker ps
CONTAINER ID        IMAGE               COMMAND                CREATED                  STATUS                  PORTS                                 NAMES
832f42819adc        nginx:1             "nginx -g 'daemon of   Less than a second ago   Up Less than a second   443/tcp, 192.168.136.136:81->80/tcp   box3/thirsty_hawking
640297cb29a7        nginx:1             "nginx -g 'daemon of   8 seconds ago            Up About a minute                                             box3/furious_heisenberg
7ecf562b1b3f        nginx:1             "nginx -g 'daemon of   13 seconds ago           Up About a minute                                             box2/ecstatic_meitner
09a92f582bc2        nginx:1             "nginx -g 'daemon of   About a minute ago       Up About a minute                                             box1/mad_goldstine

Dependency Filter

This filter co-schedules dependent containers on the same node.

Currently, dependencies are declared as follows:

• Shared volumes: --volumes-from=dependency
• Links: --link=dependency:alias
• Shared network stack: --net=container:dependency

Swarm will attempt to co-locate the dependent container on the same node. If it cannot be done (because the dependent container doesn't exist, or because the node doesn't have enough resources), it will prevent the container creation.

The combination of multiple dependencies will be honored if possible. For instance, --volumes-from=A --net=container:B will attempt to co-locate the container on the same node as A and B. If those containers are running on different nodes, Swarm will prevent you from scheduling the container.

Health Filter

This filter will prevent scheduling containers on unhealthy nodes.

Docker Swarm documentation index

• User guide
• Discovery options
• Sheduler strategies
• Swarm API