Add documentation related to oversubscribe check

docs/additional-resources/changelog.md

@@ -17,7 +17,8 @@ The following is a high-level list of notable changes made to CU Research Comput
 | Apr | `amc`, `csu`, and `rmacc` partitions removed | Affects user job submission parameters |
 | Apr | SDS production rollout at [sds.rc.colorado.edu](https://sds.rc.colorado.edu) | New CURC service providing detailed information on software available to users (caveat: some content is AI-generated) |
 | Apr | RCAMP upgrade (including new Django interface) | New interface for adding/removing users from groups/projects; Safari may no longer render interface correctly |
-| Jun | Retired `mem` QoS, introduced `mem-normal` and `mem-long` QoSes | Affects user job submission parameters |
-| Jul | Created the QoSes `gpu-normal`, `gpu-long`, and `gpu-testing`. Replaced all GPU partition QoSes with these new QoSes. Removed the `atesting_a100` and `atesting_mi100` partitions. | Affects user job submission parameters and max GPU limits per user and QoS |
-| Aug | Renamed the `amilan` partition to `acpu` and the `normal` and `long` QoSes to `cpu-normal` and `cpu-long`, respectively. Added 17 CPU-only nodes to the `acpu` partition, 8 nodes with 4 RTX Pro 6000 GPUs each to the new `artxpro6000` partition, and 8 nodes with 4 H200 GPUs each to the new `ah200` partition. |  Increased availability of CPU and GPU resources and affects user job submission parameters |
+| Jun | Retired `mem` QoS, introduced `mem-normal` and `mem-long` QoS | Affects user job submission parameters |
+| Jul | Created the QoS `gpu-normal`, `gpu-long`, and `gpu-testing`. Replaced all GPU partition QoS with these new QoS. Removed the `atesting_a100` and `atesting_mi100` partitions. | Affects user job submission parameters and max GPU limits per user and QoS |
+| Aug | Renamed the `amilan` partition to `acpu` and the `normal` and `long` QoS to `cpu-normal` and `cpu-long`, respectively. Added 17 CPU-only nodes to the `acpu` partition, 8 nodes with 4 RTX Pro 6000 GPUs each to the new `artxpro6000` partition, and 8 nodes with 4 H200 GPUs each to the new `ah200` partition. |  Increased availability of CPU and GPU resources and affects user job submission parameters |
+| Sep | Added job submission checks that ensure the `--oversubscribe` Slurm directive is only used with the `acompile` and `ahub` partitions. | Affects user job submission parameters. |
 ::::

docs/running-jobs/error-status-codes.md

@@ -78,6 +78,7 @@ documentation.](https://slurm.schedmd.com/squeue.html#lbAF)
 | 19 | The requested runtime of the `<qos>` QoS must be greater than `<N>` hour(s). | You have not requested enough time to meet the minimum required for this QoS. | Increase the time requested for the job, or switch to a different QoS that supports shorter jobs. You can find a list of valid QoS values and their associated runtime constraints in our [Alpine QoS Documentation](../clusters/alpine/alpine-hardware.md#quality-of-service-qos). |
 | 20 | When requesting resources from GPU partitions, users must specify `--gres`. | You have not specified a `GRES` using the `--gres` directive, which is required on this partition. | Specify a `GRES` using the `--gres` directive. You can find a list of Alpine `GRES`, as well as examples, in our [Alpine GRES Documentation](../clusters/alpine/alpine-hardware.md#general-resources-gres). |
 | 21 | If a list of GPU partitions is provided, then `GRES` must be set to `--gres=gpu:N`, where `N` is the number of GPUs. | You have provided an invalid `GRES` format while using a list of GPU partitions. | Use the `GRES` format `--gres=gpu:N`, where `N` is the number of GPUs. |
+| 22 | The oversubscribe Slurm directive cannot be used with the provided partition. | You have provided the `--oversubscribe` Slurm directive and it is not allowed to be used with the provided partition. | Remove the `--oversubscribe` Slurm directive or change to a compatible partition. |
 
 ## Alpine Job Submission Warning Codes