Skip to content

Commit

Permalink
scheduler-doc
Browse files Browse the repository at this point in the history
  • Loading branch information
yarongol committed Sep 25, 2024
1 parent 3f50711 commit b09b13c
Show file tree
Hide file tree
Showing 7 changed files with 115 additions and 142 deletions.
Binary file added docs/Researcher/scheduling/img/fairshare.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added docs/Researcher/scheduling/img/queue.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added docs/Researcher/scheduling/img/quota-split.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
245 changes: 109 additions & 136 deletions docs/Researcher/scheduling/the-runai-scheduler.md

Large diffs are not rendered by default.

6 changes: 3 additions & 3 deletions docs/Researcher/workloads/workload-support.md
Original file line number Diff line number Diff line change
Expand Up @@ -17,8 +17,8 @@ Different types of workloads have different levels of support. Understanding wha
| ----- | :---: | :---: | :---: | :---: | ----- |
| | Run:ai workloads | | | | Third-party workloads |
| | Training - Standard | Workspace | Inference | Training - distributed | All K8s workloads |
| [Fairness](../../Researcher/scheduling/the-runai-scheduler.md#fairness) | v | v | v | v | v |
| [Priority and preemption](../../Researcher/scheduling/the-runai-scheduler.md#allocation--preemption) | v | v | v | v | v |
| [Fairness](../../Researcher/scheduling/the-runai-scheduler.md#fairness-fair-resource-distribution) | v | v | v | v | v |
| [Priority and preemption](../../Researcher/scheduling/the-runai-scheduler.md#preemption) | v | v | v | v | v |
| [Over quota](../../Researcher/scheduling/the-runai-scheduler.md#over-quota-priority) | v | v | v | v | v |
| [Node pools](../../platform-admin/aiinitiatives/resources/node-pools.md) | v | v | v | v | v |
| Bin packing / Spread | v | v | v | v | v |
Expand All @@ -27,7 +27,7 @@ Different types of workloads have different levels of support. Understanding wha
| Node level scheduler | v | v | v | v | v |
| GPU swap | v | v | v | v | v |
| Elastic scaling | NA | NA | v | v | v |
| [Gang scheduling](../../Researcher/scheduling/the-runai-scheduler.md#distributed-training) | v | v | v | v | v |
| [Gang scheduling](../../Researcher/scheduling/the-runai-scheduler.md#gang-scheduling) | v | v | v | v | v |
| [Monitoring](../../admin/maintenance/alert-monitoring.md) | v | v | v | v | v |
| [RBAC](../../admin/authentication/authentication-overview.md#role-based-access-control-rbac-in-runai) | v | v | v | v | |
| Workload awareness | v | v | v | v | |
Expand Down
4 changes: 2 additions & 2 deletions docs/home/whats-new-2-15.md
Original file line number Diff line number Diff line change
Expand Up @@ -87,9 +87,9 @@ date: 2023-Dec-3

#### Quotas and Over-Quota

* <!-- RUN-10251/RUN-10252 - Block over-subscription of quota by Projects/Departments- -->Improved control over how over-quota is managed by adding the ability to block over-subscription of the quota in *Projects* or *Departments*. For more information, see [Limit Over-Quota](../Researcher/scheduling/the-runai-scheduler.md#limit-quota-over-or-under-subscription).
* <!-- RUN-10251/RUN-10252 - Block over-subscription of quota by Projects/Departments- -->Improved control over how over-quota is managed by adding the ability to block over-subscription of the quota in *Projects* or *Departments*. For more information, see [Limit Over-Quota](../Researcher/scheduling/the-runai-scheduler.md#over-quota).
* <!-- RUN-13167/RUN-13168 Department Over-Quota Priority behavior -->Improved the scheduler fairness for departments using the `over quota priority` switch (in Settings). When the feature flag is disabled, over-quota weights are equal to the deserved quota and any excess resources are divided in the same proportion as the in-quota resources. For more information, see [Over Quota Priority](../Researcher/scheduling/the-runai-scheduler.md#over-quota-priority).
* Added new functionality to always guarantee in-quota workloads at the expense of inter-Department fairness. Large distributed workloads from one department may preempt in-quota smaller workloads from another department. This new setting in the `RunaiConfig` file preserves in-quota workloads, even if the department quota or over-quota-fairness is not preserved. For more information, see [Scheduler Fairness](../Researcher/scheduling/the-runai-scheduler.md#fairness).
* Added new functionality to always guarantee in-quota workloads at the expense of inter-Department fairness. Large distributed workloads from one department may preempt in-quota smaller workloads from another department. This new setting in the `RunaiConfig` file preserves in-quota workloads, even if the department quota or over-quota-fairness is not preserved. For more information, see [Scheduler Fairness](../Researcher/scheduling/the-runai-scheduler.md#fairness-fair-resource-distribution).

<!-- #### Notifications
Expand Down
2 changes: 1 addition & 1 deletion graveyard/whats-new-2-14.md
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ TODO Add RBAC old--new conversion table here. -->

### Scheduler

* Added new functionality to always guarantee in-quota workloads at the expense of inter-Department fairness. Large distributed workloads from one *Department* may preempt in-quota smaller workloads from another *Department*. This new setting in the `RunaiConfig` file preserves in-quota workloads, even if the department quota/over-quota-fairness is not preserved. For more information, see [Scheduler Fairness](../Researcher/scheduling/the-runai-scheduler.md#fairness).
* Added new functionality to always guarantee in-quota workloads at the expense of inter-Department fairness. Large distributed workloads from one *Department* may preempt in-quota smaller workloads from another *Department*. This new setting in the `RunaiConfig` file preserves in-quota workloads, even if the department quota/over-quota-fairness is not preserved. For more information, see [Scheduler Fairness](../Researcher/scheduling/the-runai-scheduler.md).

### Ephemeral volumes

Expand Down

0 comments on commit b09b13c

Please sign in to comment.