basehub/binderhub: update fixme comments about resource req/lim

Author: consideRatio

helm-charts/basehub/values.yaml

@@ -129,9 +129,11 @@ jupyterhub:
       replicas: 0
     userScheduler:
       enabled: false
+      # FIXME: We should think about these resource requests/limits, see
+      #        https://github.com/2i2c-org/infrastructure/issues/2127.
+      #
       resources:
         requests:
-          # FIXME: Just unset this?
           cpu: 0.01
           memory: 64Mi
         limits:
@@ -145,15 +147,22 @@ jupyterhub:
     service:
       type: ClusterIP
     chp:
+      # FIXME: We should think about these resource requests/limits, see
+      #        https://github.com/2i2c-org/infrastructure/issues/2127.
+      #
       resources:
         requests:
-          # FIXME: We want no guarantees here!!!
-          # This is lowest possible value
           cpu: 0.01
           memory: 64Mi
         limits:
           memory: 1Gi
     traefik:
+      # FIXME: We should think about these resource requests/limits, see
+      #        https://github.com/2i2c-org/infrastructure/issues/2127.
+      #
+      #        Note if autohttps pod's aren't used anywhere by our basehub
+      #        deployments, we should simply remove this traefik configuration.
+      #
       resources:
         requests:
           memory: 64Mi
@@ -450,9 +459,11 @@ jupyterhub:
           ports:
             - port: http
               protocol: TCP
+    # FIXME: We should think about these resource requests/limits, see
+    #        https://github.com/2i2c-org/infrastructure/issues/2127.
+    #
     resources:
       requests:
-        # Very small unit, since we don't want any CPU guarantees
         cpu: 0.01
         memory: 128Mi
       limits:

helm-charts/binderhub/values.yaml

@@ -38,8 +38,16 @@ binderhub:
           hub.jupyter.org/node-purpose: core
         resources:
           requests:
-            # FIXME: We want no guarantees here!!!
-            # This is lowest possible value
+            # FIXME: These are basehub's defaults that may very well be worth
+            #        deviating from.
+            #
+            #        For example, we want to avoid the proxy pod gets evicted
+            #        if the node is under memory pressure and the proxy pod is
+            #        using more memory than requested. We also want to avoid
+            #        having the proxy pod starved of CPU by a minimal request if
+            #        other pods make significant requests and get a far greater
+            #        ratio of the node's available CPU.
+            #
             cpu: 0.01
             memory: 64Mi
           limits: