Skip to content

spec: add scheduler entity #1188

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
AkihiroSuda merged 1 commit into from
May 11, 2023
Merged

spec: add scheduler entity #1188

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 27 additions & 0 deletions config.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -291,6 +291,33 @@ For Linux-based systems, the `process` object supports the following process-spe

This is a per-process setting, where as [`disableOOMKiller`](config-linux.md#memory) is scoped for a memory cgroup.
For more information on how these two settings work together, see [the memory cgroup documentation section 10. OOM Contol][cgroup-v1-memory_2].
* **`scheduler`** (object, OPTIONAL) is an object describing the scheduler properties for the process. The `scheduler` contains the following properties:

* **`policy`** (string, REQUIRED) represents the scheduling policy. A valid list of values is:

* `SCHED_OTHER`
* `SCHED_FIFO`
* `SCHED_RR`
* `SCHED_BATCH`
* `SCHED_ISO`
* `SCHED_IDLE`
* `SCHED_DEADLINE`
Comment thread
giuseppe marked this conversation as resolved.
Copy link
Copy Markdown
Member

@utam0k utam0k Mar 21, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

SCHED_DEADLINE seems to sometimes return an EBUSY error. Do you want the error numbers to follow these when the set fails?
https://man7.org/linux/man-pages/man2/sched_setattr.2.html

There seems to be some difficulty in using SCHED_DEADLINE with cpuset. can we consider removing SCHED_DEADLINE from support?
https://www.kernel.org/doc/Documentation/scheduler/sched-deadline.txt

Copy link
Copy Markdown
Member Author

@giuseppe giuseppe Mar 21, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think it is required to support it, I am just listing the possible values for now (e.g. SCHED_ISO is not even implemented yet)

Copy link
Copy Markdown
Member

@utam0k utam0k Mar 21, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I will leave it up to you to specify about it in the oci-spec, but there are some things that probably cannot be handled. I thought users might expect to be able to use them.


* **`nice`** (int32, OPTIONAL) is the nice value for the process, affecting its priority. A lower nice value corresponds to a higher priority. If not set, the runtime must use the value 0.
* **`priority`** (int32, OPTIONAL) represents the static priority of the process, used by real-time policies like SCHED_FIFO and SCHED_RR. If not set, the runtime must use the value 0.
* **`flags`** (array of strings, OPTIONAL) is an array of strings representing scheduling flags. A valid list of values is:

* `SCHED_FLAG_RESET_ON_FORK`
* `SCHED_FLAG_RECLAIM`
* `SCHED_FLAG_DL_OVERRUN`
* `SCHED_FLAG_KEEP_POLICY`
* `SCHED_FLAG_KEEP_PARAMS`
* `SCHED_FLAG_UTIL_CLAMP_MIN`
* `SCHED_FLAG_UTIL_CLAMP_MAX`

* **`runtime`** (uint64, OPTIONAL) represents the amount of time in nanoseconds during which the process is allowed to run in a given period, used by the deadline scheduler. If not set, the runtime must use the value 0.
* **`deadline`** (uint64, OPTIONAL) represents the absolute deadline for the process to complete its execution, used by the deadline scheduler. If not set, the runtime must use the value 0.
* **`period`** (uint64, OPTIONAL) represents the length of the period in nanoseconds used for determining the process runtime, used by the deadline scheduler. If not set, the runtime must use the value 0.
* **`selinuxLabel`** (string, OPTIONAL) specifies the SELinux label for the process.
For more information about SELinux, see [SELinux documentation][selinux].

Expand Down
32 changes: 32 additions & 0 deletions schema/config-schema.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -147,6 +147,38 @@
"noNewPrivileges": {
"type": "boolean"
},
"scheduler": {
"type": "object",
"required": [
"policy"
],
"properties": {
"policy": {
"$ref": "defs-linux.json#/definitions/SchedulerPolicy"
},
"nice": {
"$ref": "defs.json#/definitions/int32"
},
"priority": {
"$ref": "defs.json#/definitions/int32"
},
"flags": {
"type": "array",
"items": {
"$ref": "defs-linux.json#/definitions/SchedulerFlag"
}
},
"runtime": {
"$ref": "defs.json#/definitions/uint64"
},
"deadline": {
"$ref": "defs.json#/definitions/uint64"
},
"period": {
"$ref": "defs.json#/definitions/uint64"
}
}
},
"rlimits": {
"type": "array",
"items": {
Expand Down
24 changes: 24 additions & 0 deletions schema/defs-linux.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -323,6 +323,30 @@
"$ref": "defs.json#/definitions/uint32"
}
}
},
"SchedulerPolicy": {
"type": "string",
"enum": [
"SCHED_OTHER",
"SCHED_FIFO",
"SCHED_RR",
"SCHED_BATCH",
"SCHED_ISO",
"SCHED_IDLE",
"SCHED_DEADLINE"
]
},
"SchedulerFlag": {
"type": "string",
"enum": [
"SCHED_FLAG_RESET_ON_FORK",
"SCHED_FLAG_RECLAIM",
"SCHED_FLAG_DL_OVERRUN",
"SCHED_FLAG_KEEP_POLICY",
"SCHED_FLAG_KEEP_PARAMS",
"SCHED_FLAG_UTIL_CLAMP_MIN",
"SCHED_FLAG_UTIL_CLAMP_MAX"
]
}
}
}
70 changes: 70 additions & 0 deletions specs-go/config.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,34 @@ type Spec struct {
ZOS *ZOS `json:"zos,omitempty" platform:"zos"`
}

// Scheduler represents the scheduling attributes for a process. It is based on
// the Linux sched_setattr(2) syscall.
type Scheduler struct {
// Policy represents the scheduling policy (e.g., SCHED_FIFO, SCHED_RR, SCHED_OTHER).
Policy LinuxSchedulerPolicy `json:"policy"`

// Nice is the nice value for the process, which affects its priority.
Nice int32 `json:"nice,omitempty"`

// Priority represents the static priority of the process.
Priority int32 `json:"priority,omitempty"`

// Flags is an array of scheduling flags.
Flags []LinuxSchedulerFlag `json:"flags,omitempty"`

// The following ones are used by the DEADLINE scheduler.

// Runtime is the amount of time in nanoseconds during which the process
// is allowed to run in a given period.
Runtime uint64 `json:"runtime,omitempty"`

// Deadline is the absolute deadline for the process to complete its execution.
Deadline uint64 `json:"deadline,omitempty"`

// Period is the length of the period in nanoseconds used for determining the process runtime.
Period uint64 `json:"period,omitempty"`
}

// Process contains information to start a specific application inside the container.
type Process struct {
// Terminal creates an interactive terminal for the container.
Expand Down Expand Up @@ -60,6 +88,8 @@ type Process struct {
ApparmorProfile string `json:"apparmorProfile,omitempty" platform:"linux"`
// Specify an oom_score_adj for the container.
OOMScoreAdj *int `json:"oomScoreAdj,omitempty" platform:"linux"`
// Scheduler specifies the scheduling attributes for a process
Scheduler *Scheduler `json:"scheduler,omitempty" platform:"linux"`
// SelinuxLabel specifies the selinux context that the container process is run as.
SelinuxLabel string `json:"selinuxLabel,omitempty" platform:"linux"`
}
Expand Down Expand Up @@ -789,3 +819,43 @@ type ZOSDevice struct {
// Gid of the device.
GID *uint32 `json:"gid,omitempty"`
}

// LinuxSchedulerPolicy represents different scheduling policies used with the Linux Scheduler
type LinuxSchedulerPolicy string

const (
// SchedOther is the default scheduling policy
SchedOther LinuxSchedulerPolicy = "SCHED_OTHER"
// SchedFIFO is the First-In-First-Out scheduling policy
SchedFIFO LinuxSchedulerPolicy = "SCHED_FIFO"
// SchedRR is the Round-Robin scheduling policy
SchedRR LinuxSchedulerPolicy = "SCHED_RR"
// SchedBatch is the Batch scheduling policy
SchedBatch LinuxSchedulerPolicy = "SCHED_BATCH"
// SchedISO is the Isolation scheduling policy
SchedISO LinuxSchedulerPolicy = "SCHED_ISO"
// SchedIdle is the Idle scheduling policy
SchedIdle LinuxSchedulerPolicy = "SCHED_IDLE"
// SchedDeadline is the Deadline scheduling policy
SchedDeadline LinuxSchedulerPolicy = "SCHED_DEADLINE"
)

// LinuxSchedulerFlag represents the flags used by the Linux Scheduler.
type LinuxSchedulerFlag string

const (
// SchedFlagResetOnFork represents the reset on fork scheduling flag
SchedFlagResetOnFork LinuxSchedulerFlag = "SCHED_FLAG_RESET_ON_FORK"
// SchedFlagReclaim represents the reclaim scheduling flag
SchedFlagReclaim LinuxSchedulerFlag = "SCHED_FLAG_RECLAIM"
// SchedFlagDLOverrun represents the deadline overrun scheduling flag
SchedFlagDLOverrun LinuxSchedulerFlag = "SCHED_FLAG_DL_OVERRUN"
// SchedFlagKeepPolicy represents the keep policy scheduling flag
SchedFlagKeepPolicy LinuxSchedulerFlag = "SCHED_FLAG_KEEP_POLICY"
// SchedFlagKeepParams represents the keep parameters scheduling flag
SchedFlagKeepParams LinuxSchedulerFlag = "SCHED_FLAG_KEEP_PARAMS"
// SchedFlagUtilClampMin represents the utilization clamp minimum scheduling flag
SchedFlagUtilClampMin LinuxSchedulerFlag = "SCHED_FLAG_UTIL_CLAMP_MIN"
// SchedFlagUtilClampMin represents the utilization clamp maximum scheduling flag
SchedFlagUtilClampMax LinuxSchedulerFlag = "SCHED_FLAG_UTIL_CLAMP_MAX"
)