Documentation
¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
Functions ¶
This section is empty.
Types ¶
type CapacityError ¶
type CapacityError interface {
// If true, wait before trying to create more instances.
IsCapacityError() bool
// If true, the condition is specific to the requested
// instance types. Wait before trying to create more
// instances of that same type.
IsInstanceTypeSpecific() bool
error
}
A CapacityError should be returned by an InstanceSet's Create method when the cloud service indicates it has insufficient capacity to create new instances -- i.e., we shouldn't retry right away.
type Driver ¶
type Driver interface {
InstanceSet(config json.RawMessage, id InstanceSetID, tags SharedResourceTags, logger logrus.FieldLogger, reg *prometheus.Registry) (InstanceSet, error)
}
A Driver returns an InstanceSet that uses the given InstanceSetID and driver-dependent configuration parameters.
If the driver creates cloud resources that aren't attached to a single VM instance (like SSH key pairs on AWS) and support tagging, they should be tagged with the provided SharedResourceTags.
The supplied id will be of the form "zzzzz-zzzzz-zzzzzzzzzzzzzzz" where each z can be any alphanum. The returned InstanceSet must use this id to tag long-lived cloud resources that it creates, and must assume control of any existing resources that are tagged with the same id. Tagging can be accomplished by including the ID in resource names, using the cloud provider's tagging feature, or any other mechanism. The tags must be visible to another instance of the same driver running on a different host.
The returned InstanceSet must not modify or delete cloud resources unless they are tagged with the given InstanceSetID or the caller (dispatcher) calls Destroy() on them. It may log a summary of untagged resources once at startup, though. Thus, two identically configured InstanceSets running on different hosts with different ids should log about the existence of each other's resources at startup, but will not interfere with each other.
The dispatcher always passes the InstanceSetID as a tag when calling Create() and Instances(), so the driver does not need to tag/filter VMs by InstanceSetID itself.
Example:
type exampleInstanceSet struct {
ownID string
AccessKey string
}
type exampleDriver struct {}
func (*exampleDriver) InstanceSet(config json.RawMessage, id cloud.InstanceSetID, tags cloud.SharedResourceTags, logger logrus.FieldLogger, reg *prometheus.Registry) (cloud.InstanceSet, error) {
var is exampleInstanceSet
if err := json.Unmarshal(config, &is); err != nil {
return nil, err
}
is.ownID = id
return &is, nil
}
func DriverFunc ¶
func DriverFunc(fn func(config json.RawMessage, id InstanceSetID, tags SharedResourceTags, logger logrus.FieldLogger, reg *prometheus.Registry) (InstanceSet, error)) Driver
DriverFunc makes a Driver using the provided function as its InstanceSet method. This is similar to http.HandlerFunc.
type Executor ¶
type Executor interface {
// Update the set of private keys used to authenticate to
// targets.
SetSigners(...ssh.Signer)
// Set the target used for subsequent command executions.
SetTarget(ExecutorTarget)
// Return the current target.
Target() ExecutorTarget
// Execute a shell command and return the resulting stdout and
// stderr. stdin can be nil.
Execute(cmd string, stdin io.Reader) (stdout, stderr []byte, err error)
}
An Executor executes commands on an ExecutorTarget.
type ExecutorTarget ¶
type ExecutorTarget interface {
// SSH server hostname or IP address, or empty string if
// unknown while instance is booting.
Address() string
// Remote username to send during SSH authentication.
RemoteUser() string
// Return nil if the given public key matches the instance's
// SSH server key. If the provided Dialer is not nil,
// VerifyHostKey can use it to make outgoing network
// connections from the instance -- e.g., to use the cloud's
// "this instance's metadata" API.
//
// Return ErrNotImplemented if no verification mechanism is
// available.
VerifyHostKey(ssh.PublicKey, *ssh.Client) error
}
An ExecutorTarget is a remote command execution service.
type Instance ¶
type Instance interface {
ExecutorTarget
// ID returns the provider's instance ID. It must be stable
// for the life of the instance.
ID() InstanceID
// String typically returns the cloud-provided instance ID.
String() string
// Cloud provider's "instance type" ID. Matches a ProviderType
// in the cluster's InstanceTypes configuration.
ProviderType() string
// Get current tags
Tags() InstanceTags
// Replace tags with the given tags
SetTags(InstanceTags) error
// Get recent price history, if available. The InstanceType is
// supplied as an argument so the driver implementation can
// account for AddedScratch cost without requesting the volume
// attachment information from the provider's API.
PriceHistory(arvados.InstanceType) []InstancePrice
// Shut down the node
Destroy() error
}
Instance is implemented by the provider-specific instance types.
type InstancePrice ¶
func NormalizePriceHistory ¶
func NormalizePriceHistory(prices []InstancePrice) []InstancePrice
NormalizePriceHistory de-duplicates and sorts instance prices, most recent first.
type InstanceSet ¶
type InstanceSet interface {
// Create a new instance with the given type, image, and
// initial set of tags. If supported by the driver, add the
// provided public key to /root/.ssh/authorized_keys.
//
// The given InitCommand should be executed on the newly
// created instance. This is optional for a driver whose
// instances' VerifyHostKey() method never returns
// ErrNotImplemented. InitCommand will be under 1 KiB.
//
// The returned error should implement RateLimitError and
// QuotaError where applicable.
Create(arvados.InstanceType, ImageID, InstanceTags, InitCommand, ssh.PublicKey) (Instance, error)
// Return all instances, including ones that are booting or
// shutting down. Optionally, filter out nodes that don't have
// all of the given InstanceTags (the caller will ignore these
// anyway).
//
// An instance returned by successive calls to Instances() may
// -- but does not need to -- be represented by the same
// Instance object each time. Thus, the caller is responsible
// for de-duplicating the returned instances by comparing the
// InstanceIDs returned by the instances' ID() methods.
Instances(InstanceTags) ([]Instance, error)
// Stop any background tasks and release other resources.
Stop()
}
An InstanceSet manages a set of VM instances created by an elastic cloud provider like AWS, GCE, or Azure.
All public methods of an InstanceSet, and all public methods of the instances it returns, are goroutine safe.
type QuotaError ¶
type QuotaError interface {
// If true, don't create more instances until some existing
// instances are destroyed. If false, don't handle the error
// as a quota error.
IsQuotaError() bool
error
}
A QuotaError should be returned by an InstanceSet when the cloud service indicates the account cannot create more VMs than already exist.
Source Files
Directories