7. Resource Constraints
7.1. Deciding Which Nodes a Resource Can Run On
Location constraints tell the cluster which nodes a resource can run on.
There are two alternative strategies. One way is to say that, by default, resources can run anywhere, and then the location constraints specify nodes that are not allowed (an opt-out cluster). The other way is to start with nothing able to run anywhere, and use location constraints to selectively enable allowed nodes (an opt-in cluster).
Whether you should choose opt-in or opt-out depends on your personal preference and the make-up of your cluster. If most of your resources can run on most of the nodes, then an opt-out arrangement is likely to result in a simpler configuration. On the other-hand, if most resources can only run on a small subset of nodes, an opt-in configuration might be simpler.
7.1.1. Location Properties
Name |
Type |
Default |
Description |
|---|---|---|---|
id |
A unique name for the constraint (required) |
||
rsc |
The name of the resource to which this constraint applies. A location
constraint must either have a |
||
rsc-pattern |
A pattern matching the names of resources to which this constraint
applies. The syntax is the same as POSIX [http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap09.html#tag_09_04]
extended regular expressions, with the addition of an initial |
||
node |
The name of the node to which this constraint applies. A location
constraint must either have a |
||
score |
Positive values indicate a preference for running the affected
resource(s) on |
||
role |
|
This is significant only for
promotable clones, is allowed only if
|
|
resource-discovery |
always |
Whether Pacemaker should perform resource discovery (that is, check whether the resource is already running) for this resource on this node. This should normally be left as the default, so that rogue instances of a service can be stopped when they are running where they are not supposed to be. However, there are two situations where disabling resource discovery is a good idea: when a service is not installed on a node, discovery might return an error (properly written OCF agents will not, so this is usually only seen with other agent types); and when Pacemaker Remote is used to scale a cluster to hundreds of nodes, limiting resource discovery to allowed nodes can significantly boost performance. Allowed values:
|
Warning
Setting resource-discovery to never or exclusive removes Pacemaker’s
ability to detect and stop unwanted instances of a service running
where it’s not supposed to be. It is up to the system administrator (you!)
to make sure that the service can never be active on nodes without
resource-discovery (such as by leaving the relevant software uninstalled).
7.1.2. Asymmetrical “Opt-In” Clusters
To create an opt-in cluster, start by preventing resources from running anywhere by default:
# crm_attribute --name symmetric-cluster --update false
Then start enabling nodes. The following fragment says that the web server prefers sles-1, the database prefers sles-2 and both can fail over to sles-3 if their most preferred node fails.
Opt-in location constraints for two resources
<constraints>
<rsc_location id="loc-1" rsc="Webserver" node="sles-1" score="200"/>
<rsc_location id="loc-2" rsc="Webserver" node="sles-3" score="0"/>
<rsc_location id="loc-3" rsc="Database" node="sles-2" score="200"/>
<rsc_location id="loc-4" rsc="Database" node="sles-3" score="0"/>
</constraints>
7.1.3. Symmetrical “Opt-Out” Clusters
To create an opt-out cluster, start by allowing resources to run anywhere by default:
# crm_attribute --name symmetric-cluster --update true
Then start disabling nodes. The following fragment is the equivalent of the above opt-in configuration.
Opt-out location constraints for two resources
<constraints>
<rsc_location id="loc-1" rsc="Webserver" node="sles-1" score="200"/>
<rsc_location id="loc-2-do-not-run" rsc="Webserver" node="sles-2" score="-INFINITY"/>
<rsc_location id="loc-3-do-not-run" rsc="Database" node="sles-1" score="-INFINITY"/>
<rsc_location id="loc-4" rsc="Database" node="sles-2" score="200"/>
</constraints>
7.1.4. What if Two Nodes Have the Same Score
If two nodes have the same score, then the cluster will choose one. This choice may seem random and may not be what was intended, however the cluster was not given enough information to know any better.
Constraints where a resource prefers two nodes equally
<constraints>
<rsc_location id="loc-1" rsc="Webserver" node="sles-1" score="INFINITY"/>
<rsc_location id="loc-2" rsc="Webserver" node="sles-2" score="INFINITY"/>
<rsc_location id="loc-3" rsc="Database" node="sles-1" score="500"/>
<rsc_location id="loc-4" rsc="Database" node="sles-2" score="300"/>
<rsc_location id="loc-5" rsc="Database" node="sles-2" score="200"/>
</constraints>
In the example above, assuming no other constraints and an inactive cluster, Webserver would probably be placed on sles-1 and Database on sles-2. It would likely have placed Webserver based on the node’s uname and Database based on the desire to spread the resource load evenly across the cluster. However other factors can also be involved in more complex configurations.
7.1.5. Specifying locations using pattern matching
A location constraint can affect all resources whose IDs match a given pattern. The following example bans resources named ip-httpd, ip-asterisk, ip-gateway, etc., from node1.
Location constraint banning all resources matching a pattern from one node
<constraints>
<rsc_location id="ban-ips-from-node1" rsc-pattern="ip-.*" node="node1" score="-INFINITY"/>
</constraints>
7.2. Specifying the Order in which Resources Should Start/Stop
Ordering constraints tell the cluster the order in which certain resource actions should occur.
Important
Ordering constraints affect only the ordering of resource actions; they do not require that the resources be placed on the same node. If you want resources to be started on the same node and in a specific order, you need both an ordering constraint and a colocation constraint (see Placing Resources Relative to other Resources), or alternatively, a group (see Groups - A Syntactic Shortcut).
7.2.1. Ordering Properties
Field |
Default |
Description |
|---|---|---|
id |
A unique name for the constraint |
|
first |
Name of the resource that the |
|
then |
Name of the dependent resource |
|
first-action |
start |
The action that the |
then-action |
value of |
The action that the |
kind |
Mandatory |
How to enforce the constraint. Allowed values:
|
symmetrical |
TRUE for |
If true, the reverse of the constraint applies for the
opposite action (for example, if B starts after A starts,
then B stops before A stops). |
Promote and demote apply to promotable
clone resources.
7.2.2. Optional and mandatory ordering
Here is an example of ordering constraints where Database must start before Webserver, and IP should start before Webserver if they both need to be started:
Optional and mandatory ordering constraints
<constraints>
<rsc_order id="order-1" first="IP" then="Webserver" kind="Optional"/>
<rsc_order id="order-2" first="Database" then="Webserver" kind="Mandatory" />
</constraints>
Because the above example lets symmetrical default to TRUE, Webserver
must be stopped before Database can be stopped, and Webserver should be
stopped before IP if they both need to be stopped.
7.2.3. Symmetric and asymmetric ordering
A mandatory symmetric ordering of “start A then start B” implies not only that the start actions must be ordered, but that B is not allowed to be active unless A is active. For example, if the ordering is added to the configuration when A is stopped (due to target-role, failure, etc.) and B is already active, then B will be stopped.
By contrast, asymmetric ordering of “start A then start B” means the stops can occur in either order, which implies that B can remain active in the same situation.
7.3. Placing Resources Relative to other Resources
Colocation constraints tell the cluster that the location of one resource depends on the location of another one.
Colocation has an important side-effect: it affects the order in which resources are assigned to a node. Think about it: You can’t place A relative to B unless you know where B is 1.
So when you are creating colocation constraints, it is important to consider whether you should colocate A with B, or B with A.
Important
Colocation constraints affect only the placement of resources; they do not require that the resources be started in a particular order. If you want resources to be started on the same node and in a specific order, you need both an ordering constraint (see Specifying the Order in which Resources Should Start/Stop) and a colocation constraint, or alternatively, a group (see Groups - A Syntactic Shortcut).
7.3.1. Colocation Properties
Field |
Default |
Description |
|---|---|---|
id |
A unique name for the constraint (required). |
|
rsc |
The name of a resource that should be located
relative to |
|
with-rsc |
The name of the resource used as the colocation
target. The cluster will decide where to put this
resource first and then decide where to put |
|
node-attribute |
#uname |
If |
score |
0 |
Positive values indicate the resources should run on
the same node. Negative values indicate the resources
should run on different nodes. Values of
+/- |
rsc-role |
Started |
If |
with-rsc-role |
Started |
If |
influence |
value of
|
Whether to consider the location preferences of
|
7.3.2. Mandatory Placement
Mandatory placement occurs when the constraint’s score is
+INFINITY or -INFINITY. In such cases, if the constraint can’t be
satisfied, then the rsc resource is not permitted to run. For
score=INFINITY, this includes cases where the with-rsc resource is
not active.
If you need resource A to always run on the same machine as resource B, you would add the following constraint:
Mandatory colocation constraint for two resources
<rsc_colocation id="colocate" rsc="A" with-rsc="B" score="INFINITY"/>
Remember, because INFINITY was used, if B can’t run on any of the cluster nodes (for whatever reason) then A will not be allowed to run. Whether A is running or not has no effect on B.
Alternatively, you may want the opposite – that A cannot
run on the same machine as B. In this case, use score="-INFINITY".
Mandatory anti-colocation constraint for two resources
<rsc_colocation id="anti-colocate" rsc="A" with-rsc="B" score="-INFINITY"/>
Again, by specifying -INFINITY, the constraint is binding. So if the only place left to run is where B already is, then A may not run anywhere.
As with INFINITY, B can run even if A is stopped. However, in this case A also can run if B is stopped, because it still meets the constraint of A and B not running on the same node.
7.3.3. Advisory Placement
If mandatory placement is about “must” and “must not”, then advisory placement is the “I’d prefer if” alternative.
For colocation constraints with scores greater than -INFINITY and less than INFINITY, the cluster will try to accommodate your wishes, but may ignore them if other factors outweigh the colocation score. Those factors might include other constraints, resource stickiness, failure thresholds, whether other resources would be prevented from being active, etc.
Advisory colocation constraint for two resources
<rsc_colocation id="colocate-maybe" rsc="A" with-rsc="B" score="500"/>
7.3.4. Colocation by Node Attribute
The node-attribute property of a colocation constraints allows you to express
the requirement, “these resources must be on similar nodes”.
As an example, imagine that you have two Storage Area Networks (SANs) that are
not controlled by the cluster, and each node is connected to one or the other.
You may have two resources r1 and r2 such that r2 needs to use the same
SAN as r1, but doesn’t necessarily have to be on the same exact node.
In such a case, you could define a node attribute named
san, with the value san1 or san2 on each node as appropriate. Then, you
could colocate r2 with r1 using node-attribute set to san.
7.3.5. Colocation Influence
By default, if A is colocated with B, the cluster will take into account A’s preferences when deciding where to place B, to maximize the chance that both resources can run.
For a detailed look at exactly how this occurs, see Colocation Explained [http://clusterlabs.org/doc/Colocation_Explained.pdf].
However, if influence is set to false in the colocation constraint,
this will happen only if B is inactive and needing to be started. If B is
already active, A’s preferences will have no effect on placing B.
An example of what effect this would have and when it would be desirable would
be a nonessential reporting tool colocated with a resource-intensive service
that takes a long time to start. If the reporting tool fails enough times to
reach its migration threshold, by default the cluster will want to move both
resources to another node if possible. Setting influence to false on
the colocation constraint would mean that the reporting tool would be stopped
in this situation instead, to avoid forcing the service to move.
The critical resource meta-attribute is a convenient way to specify the
default for all colocation constraints and groups involving a particular
resource.
Note
If a noncritical resource is a member of a group, all later members of the group will be treated as noncritical, even if they are marked as (or left to default to) critical.
7.4. Resource Sets
Resource sets allow multiple resources to be affected by a single constraint.
A set of 3 resources
<resource_set id="resource-set-example">
<resource_ref id="A"/>
<resource_ref id="B"/>
<resource_ref id="C"/>
</resource_set>
Resource sets are valid inside rsc_location, rsc_order
(see Ordering Sets of Resources), rsc_colocation
(see Colocating Sets of Resources), and rsc_ticket
(see Configuring Ticket Dependencies) constraints.
A resource set has a number of properties that can be set, though not all have an effect in all contexts.
Field |
Default |
Description |
|---|---|---|
id |
A unique name for the set (required) |
|
sequential |
true |
Whether the members of the set must be acted on in
order. Meaningful within |
require-all |
true |
Whether all members of the set must be active before
continuing. With the current implementation, the
cluster may continue even if only one member of the
set is started, but if more than one member of the set
is starting at the same time, the cluster will still
wait until all of those have started before continuing
(this may change in future versions). Meaningful
within |
role |
The constraint applies only to resource set members
that are Promotable clones in this
role. Meaningful within |
|
action |
start |
The action that applies to all members of the set.
Meaningful within |
score |
Advanced use only. Use a specific score for this set within the constraint. |
7.5. Ordering Sets of Resources
A common situation is for an administrator to create a chain of ordered resources, such as:
A chain of ordered resources
<constraints>
<rsc_order id="order-1" first="A" then="B" />
<rsc_order id="order-2" first="B" then="C" />
<rsc_order id="order-3" first="C" then="D" />
</constraints>
Visual representation of the four resources’ start order for the above constraints
7.5.1. Ordered Set
To simplify this situation, Resource Sets can be used within ordering constraints:
A chain of ordered resources expressed as a set
<constraints>
<rsc_order id="order-1">
<resource_set id="ordered-set-example" sequential="true">
<resource_ref id="A"/>
<resource_ref id="B"/>
<resource_ref id="C"/>
<resource_ref id="D"/>
</resource_set>
</rsc_order>
</constraints>
While the set-based format is not less verbose, it is significantly easier to get right and maintain.
Important
If you use a higher-level tool, pay attention to how it exposes this functionality. Depending on the tool, creating a set A B may be equivalent to A then B, or B then A.
7.5.2. Ordering Multiple Sets
The syntax can be expanded to allow sets of resources to be ordered relative to
each other, where the members of each individual set may be ordered or
unordered (controlled by the sequential property). In the example below, A
and B can both start in parallel, as can C and D, however C and
D can only start once both A and B are active.
Ordered sets of unordered resources
<constraints>
<rsc_order id="order-1">
<resource_set id="ordered-set-1" sequential="false">
<resource_ref id="A"/>
<resource_ref id="B"/>
</resource_set>
<resource_set id="ordered-set-2" sequential="false">
<resource_ref id="C"/>
<resource_ref id="D"/>
</resource_set>
</rsc_order>
</constraints>
Visual representation of the start order for two ordered sets of unordered resources
Of course either set – or both sets – of resources can also be internally
ordered (by setting sequential="true") and there is no limit to the number
of sets that can be specified.
Advanced use of set ordering - Three ordered sets, two of which are internally unordered
<constraints>
<rsc_order id="order-1">
<resource_set id="ordered-set-1" sequential="false">
<resource_ref id="A"/>
<resource_ref id="B"/>
</resource_set>
<resource_set id="ordered-set-2" sequential="true">
<resource_ref id="C"/>
<resource_ref id="D"/>
</resource_set>
<resource_set id="ordered-set-3" sequential="false">
<resource_ref id="E"/>
<resource_ref id="F"/>
</resource_set>
</rsc_order>
</constraints>
Visual representation of the start order for the three sets defined above
Important
An ordered set with sequential=false makes sense only if there is another
set in the constraint. Otherwise, the constraint has no effect.
7.5.3. Resource Set OR Logic
The unordered set logic discussed so far has all been “AND” logic. To illustrate this take the 3 resource set figure in the previous section. Those sets can be expressed, (A and B) then (C) then (D) then (E and F).
Say for example we want to change the first set, (A and B), to use “OR” logic
so the sets look like this: (A or B) then (C) then (D) then (E and F). This
functionality can be achieved through the use of the require-all option.
This option defaults to TRUE which is why the “AND” logic is used by default.
Setting require-all=false means only one resource in the set needs to be
started before continuing on to the next set.
Resource Set “OR” logic: Three ordered sets, where the first set is internally unordered with “OR” logic
<constraints>
<rsc_order id="order-1">
<resource_set id="ordered-set-1" sequential="false" require-all="false">
<resource_ref id="A"/>
<resource_ref id="B"/>
</resource_set>
<resource_set id="ordered-set-2" sequential="true">
<resource_ref id="C"/>
<resource_ref id="D"/>
</resource_set>
<resource_set id="ordered-set-3" sequential="false">
<resource_ref id="E"/>
<resource_ref id="F"/>
</resource_set>
</rsc_order>
</constraints>
Important
An ordered set with require-all=false makes sense only in conjunction with
sequential=false. Think of it like this: sequential=false modifies the set
to be an unordered set using “AND” logic by default, and adding
require-all=false flips the unordered set’s “AND” logic to “OR” logic.
7.6. Colocating Sets of Resources
Another common situation is for an administrator to create a set of colocated resources.
The simplest way to do this is to define a resource group (see Groups - A Syntactic Shortcut), but that cannot always accurately express the desired relationships. For example, maybe the resources do not need to be ordered.
Another way would be to define each relationship as an individual constraint, but that causes a difficult-to-follow constraint explosion as the number of resources and combinations grow.
Colocation chain as individual constraints, where A is placed first, then B, then C, then D
<constraints>
<rsc_colocation id="coloc-1" rsc="D" with-rsc="C" score="INFINITY"/>
<rsc_colocation id="coloc-2" rsc="C" with-rsc="B" score="INFINITY"/>
<rsc_colocation id="coloc-3" rsc="B" with-rsc="A" score="INFINITY"/>
</constraints>
To express complicated relationships with a simplified syntax 2, resource sets can be used within colocation constraints.
Equivalent colocation chain expressed using resource_set
<constraints>
<rsc_colocation id="coloc-1" score="INFINITY" >
<resource_set id="colocated-set-example" sequential="true">
<resource_ref id="A"/>
<resource_ref id="B"/>
<resource_ref id="C"/>
<resource_ref id="D"/>
</resource_set>
</rsc_colocation>
</constraints>
Note
Within a resource_set, the resources are listed in the order they are
placed, which is the reverse of the order in which they are colocated.
In the above example, resource A is placed before resource B, which is
the same as saying resource B is colocated with resource A.
As with individual constraints, a resource that can’t be active prevents any resource that must be colocated with it from being active. In both of the two previous examples, if B is unable to run, then both C and by inference D must remain stopped.
Important
If you use a higher-level tool, pay attention to how it exposes this functionality. Depending on the tool, creating a set A B may be equivalent to A with B, or B with A.
Resource sets can also be used to tell the cluster that entire sets of
resources must be colocated relative to each other, while the individual
members within any one set may or may not be colocated relative to each other
(determined by the set’s sequential property).
In the following example, resources B, C, and D will each be colocated with A (which will be placed first). A must be able to run in order for any of the resources to run, but any of B, C, or D may be stopped without affecting any of the others.
Using colocated sets to specify a shared dependency
<constraints>
<rsc_colocation id="coloc-1" score="INFINITY" >
<resource_set id="colocated-set-2" sequential="false">
<resource_ref id="B"/>
<resource_ref id="C"/>
<resource_ref id="D"/>
</resource_set>
<resource_set id="colocated-set-1" sequential="true">
<resource_ref id="A"/>
</resource_set>
</rsc_colocation>
</constraints>
Note
Pay close attention to the order in which resources and sets are listed. While the members of any one sequential set are placed first to last (i.e., the colocation dependency is last with first), multiple sets are placed last to first (i.e. the colocation dependency is first with last).
Important
A colocated set with sequential="false" makes sense only if there is
another set in the constraint. Otherwise, the constraint has no effect.
There is no inherent limit to the number and size of the sets used.
The only thing that matters is that in order for any member of one set
in the constraint to be active, all members of sets listed after it must also
be active (and naturally on the same node); and if a set has sequential="true",
then in order for one member of that set to be active, all members listed
before it must also be active.
If desired, you can restrict the dependency to instances of promotable clone
resources that are in a specific role, using the set’s role property.
Colocation in which the members of the middle set have no interdependencies, and the last set listed applies only to promoted instances
<constraints>
<rsc_colocation id="coloc-1" score="INFINITY" >
<resource_set id="colocated-set-1" sequential="true">
<resource_ref id="F"/>
<resource_ref id="G"/>
</resource_set>
<resource_set id="colocated-set-2" sequential="false">
<resource_ref id="C"/>
<resource_ref id="D"/>
<resource_ref id="E"/>
</resource_set>
<resource_set id="colocated-set-3" sequential="true" role="Promoted">
<resource_ref id="A"/>
<resource_ref id="B"/>
</resource_set>
</rsc_colocation>
</constraints>
Visual representation of the above example (resources are placed from left to right)
Note
Unlike ordered sets, colocated sets do not use the require-all option.
7.7. External Resource Dependencies
Sometimes, a resource will depend on services that are not managed by the cluster. An example might be a resource that requires a file system that is not managed by the cluster but mounted by systemd at boot time.
To accommodate this, the pacemaker systemd service depends on a normally empty
target called resource-agents-deps.target. The system administrator may
create a unit drop-in for that target specifying the dependencies, to ensure
that the services are started before Pacemaker starts and stopped after
Pacemaker stops.
Typically, this is accomplished by placing a unit file in the
/etc/systemd/system/resource-agents-deps.target.d directory, with directives
such as Requires and After specifying the dependencies as needed.