Document CPU/RAM quotas (#2991)

This PR adds documentation on CPU/RAM quotas introduced in
WATonomous/infra-config#2619

components/quota-table.tsx

@@ -97,4 +97,34 @@ export function NodeLocalQuotaTable({
             <TableBody>{rows}</TableBody>
         </Table>
     )
-}
+}
+
+export function CPURAMQuotaTable({
+    className = "",
+}: {
+    className?: string
+}) {
+    const rows = []
+    for (const machine of machineInfo.machines.dev_vms) {
+        rows.push(
+            <TableRow>
+                <TableCell>{machine.name}</TableCell>
+                <TableCell className='text-center'>{machine.cpu_quota}</TableCell>
+                <TableCell className='text-center'>{machine.memory_quota}</TableCell>
+            </TableRow>
+        )
+    }
+
+    return (
+        <Table className={className}>
+            <TableHeader>
+                <TableRow>
+                    <TableHead>Node</TableHead>
+                    <TableHead className='text-center'>CPU Quota (% of 1 core)</TableHead>
+                    <TableHead className='text-center'>Memory Quota (bytes)</TableHead>
+                </TableRow>
+            </TableHeader>
+            <TableBody>{rows}</TableBody>
+        </Table>
+    )
+}

pages/docs/compute-cluster/machine-usage-guide.mdx

@@ -122,19 +122,6 @@ Examples of software that we do not install:
 
 If there is a piece of software that you think should be installed on the machines, please reach out to a WATcloud team member.
 
-### `OMP_NUM_THREADS`
-
-On general-use machines, we set the `OMP_NUM_THREADS` environment variable to `1` by default. This is to prevent system overload when running
-programs that use OpenMP[^openmp]. This default is consistent with the
-[default behaviour of PyTorch Distributed](https://github.com/pytorch/pytorch/blob/4b494d075093096d822b9d614e2719a0e821c6af/torch/distributed/run.py#L758-L763).
-If you'd like to change this default, simply set the `OMP_NUM_THREADS` environment variable to the desired value, for example:
-
-```bash copy
-export OMP_NUM_THREADS=4
-```
-
-[^openmp]: [OpenMP](https://www.openmp.org/) is an API that supports multi-platform shared memory multiprocessing programming in C, C++, and Fortran.
-
 ## Maintenance and Outages
 
 We try to keep the machines in the cluster up and running at all times. However, we do need to perform regular maintenance to keep the machines

pages/docs/compute-cluster/quotas.mdx

@@ -1,3 +1,5 @@
+import { GlobalQuotaTable, NodeLocalQuotaTable, CPURAMQuotaTable } from '@/components/quota-table'
+
 # Quotas
 
 To ensure that everyone has a fair share of resources, we enforce a set of quotas in the cluster.
@@ -18,8 +20,6 @@ To ensure that everyone has a fair share of resources, we enforce a set of quota
 
 This section lists the default per-user disk quotas. These quotas are subject to change as we learn more about the usage patterns.
 
-import { GlobalQuotaTable, NodeLocalQuotaTable } from '@/components/quota-table'
-
 #### Global disk quotas
 
 Global disk quotas are quotas on filesystems that are shared across all nodes.
@@ -122,6 +122,15 @@ profile using the [Profile Editor](../utilities/profile-editor).
 
 [^watcloud-contact]: Your WATcloud contact is the person as described in the [Getting Access](./getting-access#determine-your-watcloud-contact) section.
 
+## CPU and memory quotas
+
+On general-use machines, per-user CPU and memory quotas are enforced to ensure fair resource sharing.
+
+<CPURAMQuotaTable className="mt-4" />
+
+Unlike disk quotas, we don't allow users to request CPU and memory quota increases. Please use [SLURM](./slurm) to run resource-intensive jobs.
+
+
 {
 // Separate footnotes from the main content
 }

scripts/generate-machine-info.py

@@ -166,6 +166,7 @@ def generate_fixtures(data_path):
         }
 
         if "login_nodes" in group_names:
+            login_nodes_config = get_group_config(host, "login_nodes")
             properties.update({
                 "cpu_info": get_cpu_info(data_path, name),
                 "memory_info": get_memory_info(data_path, name),
@@ -174,6 +175,8 @@ def generate_fixtures(data_path):
                 "lsb_release_info": get_lsb_release_info(data_path, name),
                 "ssh_host_keys": get_file_lines(data_path, name, "ssh-host-keys.log"),
                 "mounts_with_quotas": get_mounts_with_quotas(host),
+                "cpu_quota": login_nodes_config.get("cpu_quota"),
+                "memory_quota": login_nodes_config.get("memory_max"),
             })
             dev_vms.append(properties)
         if "slurmd_nodes" in group_names: