package peloton.api.v0.task
Get desktop application:
View/edit binary Protocol Buffers messages
service TaskManager
task.proto:684* Task manager interface
rpc Get (, )
task.proto:687Get the info of a task in job.
message
task.proto:784DEPRECATED by peloton.api.v0.task.svc.GetTaskRequest.
optional jobId = 1
uint32 instanceId = 2
message
task.proto:790DEPRECATED by peloton.api.v0.task.svc.GetTaskResponse.
optional result = 1
DEPRECATED by repeated TaskInfo
optional notFound = 2
optional outOfRange = 3
repeated results = 4
Returns all active and completed tasks of the given instance.
rpc List (, )
task.proto:690List all task info in a job.
message
task.proto:866DEPRECATED by peloton.api.v0.task.svc.ListTasksRequest.
optional jobId = 1
optional range = 2
message
task.proto:872DEPRECATED by peloton.api.v0.task.svc.ListTasksResponse.
optional result = 1
optional notFound = 2
rpc Start (, )
task.proto:694Start a set of tasks for a job. Will be no-op for tasks that are currently running.
message
task.proto:881DEPRECATED by peloton.api.v0.task.svc.StartTasksRequest.
optional jobId = 1
repeated ranges = 2
message
task.proto:887DEPRECATED by peloton.api.v0.task.svc.StartTasksResponse.
optional error = 1
repeated uint32 startedInstanceIds = 2
repeated uint32 invalidInstanceIds = 3
rpc Stop (, )
task.proto:698Stop a set of tasks for a job. Will be no-op for tasks that are currently stopped.
message
task.proto:909DEPRECATED by peloton.api.v0.task.svc.StopTasksRequest. If no ranges specified, then stop all the tasks in the job.
optional jobId = 1
repeated ranges = 2
message
task.proto:920DEPRECATED by peloton.api.v0.task.svc.StopTasksResponse.
optional error = 1
repeated uint32 stoppedInstanceIds = 2
repeated uint32 invalidInstanceIds = 3
rpc Restart (, )
task.proto:702Restart a set of tasks for a job. Will start tasks that are currently stopped.
message
task.proto:933DEPRECATED by peloton.api.v0.task.svc.RestartTasksRequest.
optional jobId = 1
repeated ranges = 2
message
task.proto:939DEPRECATED by peloton.api.v0.task.svc.RestartTasksResponse.
optional notFound = 1
optional outOfRange = 2
rpc Query (, )
task.proto:705Query task info in a job, using a set of filters.
message
task.proto:945DEPRECATED by peloton.api.v0.task.svc.QueryTasksRequest.
optional jobId = 1
optional spec = 2
message
task.proto:952DEPRECATED by peloton.api.v0.task.svc.QueryTasksResponse.
optional error = 1
repeated records = 2
optional pagination = 3
rpc BrowseSandbox (, )
task.proto:708BrowseSandbox returns list of file paths inside sandbox.
message
task.proto:744DEPRECATED by peloton.api.v0.task.svc.BrowseSandboxRequest.
optional jobId = 1
uint32 instanceId = 2
string taskId = 3
Get the sandbox path of a particular task of an instance. This should be set to the mesos task id in the runtime of the task for which the sandbox is being requested. If not provided, the path of the latest task is returned.
message
task.proto:755DEPRECATED by peloton.api.v0.task.svc.BrowseSandboxResponse.
optional error = 1
string hostname = 2
string port = 3
repeated string paths = 4
string mesosMasterHostname = 5
Mesos Master hostname and port.
string mesosMasterPort = 6
rpc Refresh (, )
task.proto:712Debug only method. Allows user to load task runtime state from DB and re-execute the action associated with current state.
message
task.proto:965DEPRECATED by peloton.api.v0.task.svc.RefreshTasksRequest.
optional jobId = 1
optional range = 2
message
task.proto:971DEPRECATED by peloton.api.v0.task.svc.RefreshTasksResponse.
(message has no fields)
rpc GetCache (, )
task.proto:715Debug only method. Get the cache of a task stored in Peloton.
message
task.proto:974DEPRECATED by peloton.api.task.svc.GetTaskCacheRequest.
optional jobId = 1
The job ID to look up the task.
uint32 instanceId = 2
The instance ID of the task to get.
message
task.proto:983DEPRECATED by peloton.api.task.svc.GetTaskCacheResponse.
optional runtime = 1
The task runtime of the task.
repeated labels = 2
The labels of the task.
rpc GetPodEvents (, )
task.proto:719GetPodEvents returns pod events (state transition for a pod), in reverse chronological order. pod is singular instance of a Peloton job.
message
task.proto:802* Request message for TaskService.GetPodEvents method.
optional jobId = 1
The job ID of the task
uint32 instanceId = 2
The instance ID of the task
uint64 limit = 3
Defines the number of unique run ids for which the pod events will be returned. If a specific run id is requested, using runId, then this defaults to 1. Defaults to 10 otherwise.
string runId = 4
Unique identifier to fetch pod events of an instance for a particular run. This is an optional parameter, if unset limit number of run ids worth of pod events will be returned.
message
task.proto:827* Response message for TaskService.GetPodEvents method. Return errors: INTERNAL: if failed to get task events for internal errors.
repeated result = 1
optional error = 2
rpc DeletePodEvents (, )
task.proto:724DeletePodEvents, deletes the pod events for provided request, which is for a jobID + instanceID + less than equal to runID. Response will be successful or error on unable to delete events for input.
message
task.proto:840* Request message for TaskService.DeletePodEvents method.
optional jobId = 1
The job ID of the task
uint32 instanceId = 2
The instance ID of the task
uint64 runId = 3
deletes the run less than equal to runId.
message
task.proto:857* Response message for TaskService.DeletePodEvents method. Return errors: INTERNAL: if failed to delete task events for internal errors.
(message has no fields)
message
task.proto:154* AndConstraint represents a logical 'and' of constraints.
Used in:
repeated constraints = 1
message
task.proto:731* Failures for browsing sandbox files requests to mesos call.
Used in:
string message = 1
message
task.proto:756Used in:
optional notFound = 1
optional outOfRange = 2
optional notRunning = 3
optional failure = 4
message
task.proto:135* Constraint represents a host label constraint or a related tasks label constraint. This is used to require that a host have certain label constraints or to require that the tasks already running on the host have certain label constraints.
Used in: , , , ,
Constraint.Type type = 1
optional labelConstraint = 2
optional andConstraint = 3
optional orConstraint = 4
Used in:
UNKNOWN_CONSTRAINT = 0
Reserved for compatibility.
LABEL_CONSTRAINT = 1
AND_CONSTRAINT = 2
OR_CONSTRAINT = 3
message
task.proto:830Used in:
string message = 1
message
task.proto:43* Health check configuration for a task
Used in:
bool enabled = 1
Whether the health check is enabled.
uint32 initialIntervalSecs = 2
Start time wait in seconds. Zero or empty value would use default value of 15 from Mesos.
uint32 intervalSecs = 3
Interval in seconds between two health checks. Zero or empty value would use default value of 10 from Mesos.
uint32 maxConsecutiveFailures = 4
Max number of consecutive failures before failing health check. Zero or empty value would use default value of 3 from Mesos.
uint32 timeoutSecs = 5
Health check command timeout in seconds. Zero or empty value would use default value of 20 from Mesos.
HealthCheckConfig.Type type = 6
optional commandCheck = 7
Only applicable when type is `COMMAND`.
optional httpCheck = 8
Only applicable when type is 'HTTP'.
message
task.proto:78Used in:
string command = 1
Health check command to be executed. Note that this command by default inherits all environment varibles from the task it's monitoring, unless `unshare_environments` is set to true.
bool unshareEnvironments = 2
If set, this check will not share the environment variables of the task.
message
task.proto:88HTTP health check to be executed. Sends a GET request to scheme://<host>:port/path. Host is not configurable and is resolved automatically.
Used in:
string scheme = 1
Currently http and https are supported.
uint32 port = 2
Port to send the HTTP GET.
string path = 3
The request path.
Used in:
UNKNOWN = 0
Reserved for future compatibility of new types.
COMMAND = 1
Command line based health check
HTTP = 2
HTTP endpoint based health check
GRPC = 3
GRPC endpoint based health check
* HealthState is the health check state of a task
Used in: ,
INVALID = 0
Default value.
DISABLED = 1
If the health check config is not enabled in the task config, the initial value of the health state is DISABLED.
HEALTH_UNKNOWN = 2
If the health check config is enabled in the task config, the initial value of the health state is HEALTH_UNKNOWN.
HEALTHY = 3
In a Mesos event, If the healthy field is true and the reason field is REASON_TASK_HEALTH_CHECK_STATUS_UPDATED the health state of the task is HEALTHY
UNHEALTHY = 4
In a Mesos event, If the healthy field is false and the reason field is REASON_TASK_HEALTH_CHECK_STATUS_UPDATED the health state of the task is UNHEALTHY
message
task.proto:774DEPRECATED by google.rpc.OUT_OF_RANGE error.
Used in: , , , ,
optional jobId = 1
Entity ID of the job
uint32 instanceCount = 2
Instance count of the job
message
task.proto:554* Task InstanceID range [from, to)
Used in: , , , , , , , , , , , , , , ,
uint32 from = 1
uint32 to = 2
message
task.proto:169* LabelConstraint represents a constraint on the number of occurrences of a given label from the set of host labels or task labels present on the host.
Used in:
LabelConstraint.Kind kind = 1
Determines which labels the constraint should apply to.
LabelConstraint.Condition condition = 2
Determines which constraint there should be on the number of occurrences of the label.
optional label = 3
The label which this defines a constraint on: For Kind == HOST, each attribute on Mesos agent is transformed to a label, with `hostname` as a special label which is always inferred from agent hostname and set.
uint32 requirement = 4
A limit on the number of occurrences of the label.
* Condition represents a constraint on the number of occurrences of the label.
Used in:
CONDITION_UNKNOWN = 0
Reserved for compatibility.
CONDITION_LESS_THAN = 1
CONDITION_EQUAL = 2
CONDITION_GREATER_THAN = 3
* Kind represents whatever the constraint applies to the labels on the host or to the labels of the tasks that are located on the host.
Used in:
UNKNOWN = 0
Reserved for compatibility.
TASK = 1
HOST = 2
message
task.proto:873Used in:
map<uint32, > value = 1
message
task.proto:161* OrConstraint represents a logical 'or' of constraints.
Used in:
repeated constraints = 1
message
task.proto:245* Persistent volume configuration for a task.
Used in:
string containerPath = 1
Volume mount path inside container.
uint32 sizeMB = 2
Volume size in MB.
message
task.proto:638* Pod event of a Peloton pod instance.
Used in:
optional taskId = 1
The task ID of the task event.
string actualState = 2
Actual state of an instance
string goalState = 3
Goal State of an instance
string timestamp = 4
The time when the event was created. The time is represented in RFC3339 form with UTC timezone.
uint64 configVersion = 5
The config version currently used by the runtime.
uint64 desiredConfigVersion = 6
The desired config version that should be used by the runtime.
string agentID = 7
The agentID for the task
string hostname = 8
The host on which the task is running
string message = 9
Short human friendly message explaining state.
string reason = 10
The short reason for the task event
optional prevTaskId = 11
The previous task ID of the pod event.
string healthy = 12
The health check result of the task
optional desriedTaskId = 13
The desired mesos task ID of the task event.
message
task.proto:116* Network port configuration for a task
Used in:
string name = 1
Name of the network port, e.g. http, tchannel. Required field.
uint32 value = 2
Static port number if any. If unset, will be dynamically allocated by the scheduler
string envName = 3
Environment variable name to be exported when running a task for this port. Required field for dynamic port.
message
task.proto:218* Preemption policy for a task
Used in:
PreemptionPolicy.Type type = 1
bool killOnPreempt = 2
This policy defines if the task should be restarted after it is preempted. If set to true the task will not be rescheduled after it is preempted. If set to false the task will be rescheduled. Defaults to false This only takes effect if the task is preemptible.
* Type represents whether the instance is preemptible. If so, it might be scheduled using elastic resources from other resource pools and subject to preemption when the demands of other resource pools increase. If it is non-preemptible it will be scheduled using the reserved resources only. This can override the preemptible sla configuration at the job level.
Used in:
TYPE_INVALID = 0
TYPE_PREEMPTIBLE = 1
TYPE_NON_PREEMPTIBLE = 2
message
task.proto:953Used in:
optional notFound = 1
message
task.proto:565* QuerySpec specifies the list of query criteria for tasks. All indexed fields should be part of this message. And all fields in this message have to be indexed too.
Used in: ,
optional pagination = 1
DEPRECATED: use QueryJobRequest.pagination instead. The spec of how to do pagination for the query results.
repeated TaskState taskStates = 2
List of task states to query the tasks. Will match all tasks if the list is empty.
repeated string names = 3
List of task names to query the tasks. Will match all names if the list is empty.
repeated string hosts = 4
List of hosts to query the tasks. Will match all hosts if the list is empty.
message
task.proto:21* Resource configuration for a task.
Used in: , , ,
double cpuLimit = 1
CPU limit in number of CPU cores
double memLimitMb = 2
Memory limit in MB
double diskLimitMb = 3
Disk limit in MB
uint32 fdLimit = 4
File descriptor limit
double gpuLimit = 5
GPU limit in number of GPUs
message
task.proto:208* Restart policy for a task.
Used in:
uint32 maxFailures = 1
Max number of task failures can occur before giving up scheduling retry, no backoff for now. Default 0 means no retry on failures.
message
task.proto:451* Runtime info of an task instance in a Job
Used in: , ,
TaskState state = 1
Runtime status of the task
optional mesosTaskId = 2
The mesos task ID for this instance
string startTime = 3
The time when the instance starts to run. Will be unset if the instance hasn't started running yet. The time is represented in RFC3339 form with UTC timezone.
string completionTime = 4
The time when the instance is completed. Will be unset if the instance hasn't completed yet. The time is represented in RFC3339 form with UTC timezone.
string host = 5
The name of the host where the instance is running
map<string, uint32> ports = 6
Dynamic ports reserved on the host while this instance is running
TaskState goalState = 7
The desired state of the task which should be eventually reached by the system.
string message = 8
The message that explains the current state of a task such as why the task is failed. Only track the latest one if the task has been retried and failed multiple times.
string reason = 9
The reason that explains the current state of a task. Only track the latest one if the task has been retried and failed multiple times. See Mesos TaskStatus.Reason for more details.
uint32 failureCount = 10
The number of times the task has failed after retries.
optional volumeID = 11
persistent volume id
uint64 configVersion = 12
The config version currently used by the runtime.
uint64 desiredConfigVersion = 13
The desired config version that should be used by the runtime.
optional agentID = 14
the id of mesos agent on the host to be launched.
optional revision = 15
Revision of the current task info.
optional prevMesosTaskId = 16
The mesos task id of the previous task run of the instance.
map<string, double> resourceUsage = 17
The resource usage for this task. The map key is each resource kind in string format and the map value is the number of unit-seconds of that resource used by the job. Example: if a task that uses 1 CPU and finishes in 10 seconds, this map will contain <"cpu":10>
HealthState healthy = 18
The result of the health check
optional desiredMesosTaskId = 19
The desired mesos task ID for this instance
optional terminationStatus = 20
Termination status of the task. Set only if the task is in a non-successful terminal state such as KILLED or FAILED.
string desiredHost = 21
The name of the host where the instance should be running on upon restart. It is used for best effort in-place update/restart.
message
task.proto:888Used in:
optional notFound = 1
optional outOfRange = 2
optional failure = 3
message
task.proto:921Used in:
optional notFound = 1
optional outOfRange = 2
optional updateError = 3
message
task.proto:258* Task configuration for a given job instance Note that only add string/slice/ptr type into TaskConfig directly due to the limitation of go reflection inside our task specific config logic.
Used in: , ,
string name = 1
Name of the task
repeated labels = 2
List of user-defined labels for the task
optional resource = 3
Resource config of the task
optional container = 4
Container config of the task.
optional command = 5
Command line config of the task
optional executor = 15
Custom executor config of the task.
optional healthCheck = 6
Health check config of the task
repeated ports = 7
List of network ports to be allocated for the task
optional constraint = 8
Constraint on the attributes of the host or labels on tasks on the host that this task should run on. Use `AndConstraint`/`OrConstraint` to compose multiple constraints if necessary.
optional restartPolicy = 9
Task restart policy on failures
optional volume = 10
Persistent volume config of the task.
optional preemptionPolicy = 11
Preemption policy of the task
bool controller = 12
Whether this is a controller task. A controller is a special batch task which controls other tasks inside a job. E.g. spark driver tasks in a spark job will be a controller task.
uint32 killGracePeriodSeconds = 13
This is used to set the amount of time between when the executor sends the SIGTERM message to gracefully terminate a task and when it kills it by sending SIGKILL. If you do not set the grace period duration the default is 30 seconds.
bool revocable = 14
Whether the instance is revocable. If so, it might be scheduled using revocable resources and subject to preemption when there is resource contention on the host. This can override the revocable configuration at the job level.
message
task.proto:588* Task event of a Peloton task instance.
Used in:
optional taskId = 1
The task ID of the task event.
TaskState state = 2
The task state of the task event.
string message = 3
Short human friendly message explaining state.
string timestamp = 4
The time when the event was created. The time is represented in RFC3339 form with UTC timezone.
TaskEvent.Source source = 5
The source that generated the task event.
string hostname = 6
The host on which the task is running
string reason = 7
The short reason for the task event
string agentId = 8
The agentID for the task
optional prevTaskId = 9
The previous mesos task ID of the task event.
HealthState healthy = 10
The health check result of the task
optional desiredTaskId = 11
The desired mesos task ID of the task event.
Describes the source of the task event
Used in:
SOURCE_UNKNOWN = 0
SOURCE_JOBMGR = 1
SOURCE_RESMGR = 2
SOURCE_HOSTMGR = 3
message
task.proto:861DEPRECATED by google.rpc.INTERNAL error.
string message = 1
message
task.proto:534* Info of a task instance in a Job
Used in: , , , , , , ,
uint32 instanceId = 1
The numerical ID assigned to this instance. Instance IDs must be unique and contiguous within a job. The ID is in the range of [0, N-1] for a job with instance count of N.
optional jobId = 2
Job ID of the task
optional config = 3
Configuration of the task
optional runtime = 4
Runtime info of the instance
message
task.proto:739* Task not running error.
Used in:
string message = 1
message
task.proto:903* Error when TaskStart failed.
Used in:
string message = 1
* Runtime states of a task instance
Used in: , , ,
UNKNOWN = 0
Reserved for future compatibility of new states.
INITIALIZED = 1
The task is being initialized
PENDING = 2
The task is pending and waiting for resources
READY = 3
The task has been allocated with resources and ready for placement
PLACING = 4
The task is being placed to a host based on its resource requirements and constraints
PLACED = 5
The task has been assigned to a host matching the resource requirements and constraints
LAUNCHING = 6
The task is taken from resmgr to be launched
LAUNCHED = 15
The task is being launched in Job manager TODO: We need to correct the numbering
STARTING = 7
The task is being started by Mesos agent
RUNNING = 8
The task is running on a Mesos host
SUCCEEDED = 9
The task terminated with an exit code of zero
FAILED = 10
The task terminated with a non-zero exit code
LOST = 11
The task is lost
PREEMPTING = 12
The task is being preempted by another one on the node
KILLING = 13
The task is being killed
KILLED = 14
Execution of the task was terminated by the system
DELETED = 16
The task is to be deleted after termination
RESERVED = 17
Placement engine is reserving host for the task
message
task.proto:915DEPRECATED by google.rpc.INTERNAL error.
Used in:
string message = 1
message
task.proto:406TerminationStatus contains details about termination of a task. It mainly contains Peloton-specific reasons for termination.
Used in:
TerminationStatus.Reason reason = 1
Reason for termination.
uint32 exit_code = 2
If non-zero, exit status when the container terminated.
string signal = 3
Name of signal received by the container when it terminated.
Reason lists various causes for a task termination
Used in:
TERMINATION_STATUS_REASON_INVALID = 0
Default value.
TERMINATION_STATUS_REASON_KILLED_ON_REQUEST = 1
Task was killed because a stop request was received from a client.
TERMINATION_STATUS_REASON_FAILED = 2
Task failed. See also TerminationStatus.exit_code, TerminationStatus.signal and ContainerStatus.message.
TERMINATION_STATUS_REASON_KILLED_HOST_MAINTENANCE = 3
Task was killed to put the host in to maintenance.
TERMINATION_STATUS_REASON_PREEMPTED_RESOURCES = 4
Task was killed to reclaim resources allocated to it.
TERMINATION_STATUS_REASON_DEADLINE_TIMEOUT_EXCEEDED = 5
Task was killed due to deadline tracker exceeding task timeout
TERMINATION_STATUS_REASON_KILLED_FOR_UPDATE = 6
Task was killed for update
TERMINATION_STATUS_REASON_KILLED_FOR_RESTART = 7
Task was killed for restart
TERMINATION_STATUS_REASON_KILLED_FOR_SLA_AWARE_RESTART = 8
Task was killed for sla aware restart