You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: CHANGELOG.md
+4-1Lines changed: 4 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -14,7 +14,10 @@ The automatic allocator has been finally reimplemented, and is now much better:
14
14
* It now uses information from the scheduler to determine how many allocations to spawn, and thus it can react to the
15
15
current computational load much more accurately. It should also be less "eager".
16
16
* It properly supports multi-node tasks.
17
-
* It considers computational load across all allocation queues (before, each queue was treated separately, which led to creating too many submissions).
17
+
* It considers computational load across all allocation queues (before, each queue was treated separately, which led to
18
+
creating too many submissions).
19
+
* It now exposes a `min-utilization` parameter, which can be used to avoid spawning an allocation that couldn't be utilized
20
+
enough.
18
21
19
22
As this is a large behavioral change, we would be happy to hear your feedback!
Copy file name to clipboardExpand all lines: docs/deployment/allocation.md
+9Lines changed: 9 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -107,6 +107,15 @@ Maximum number of workers that can be queued or running in the allocation queue.
107
107
limited by the manager (PBS/Slurm), but you can use this parameter to make the limit smaller, for example if you also want
108
108
to manage allocations outside HyperQueue.
109
109
110
+
#### Minimal utilization
111
+
Format: `--min-utilization <ratio>`
112
+
113
+
Minimal utilization determines how could the scheduler utilize workers from submitted allocations. If the schedules thinks that it can make use of `N%` of worker resources in a single allocation of this queue, `min-utilization` has to be at least `N`, otherwise the allocation will not be created.
114
+
115
+
It has to be a floating point number between 0.0 and 1.0.
116
+
117
+
The default minimal utilization is `0`, which means that an allocation will be created if the scheduler thinks that it can use any (non-zero) amount of resources of worker(s) in the allocation.
118
+
110
119
#### Worker resources
111
120
You can specify [CPU](../jobs/cresources.md) and [generic](../jobs/resources.md) resources of workers spawned by the
112
121
allocation queue. The name and syntax of these parameters is the same as when you create a
0 commit comments