rearrange vars file

deployment-checklist.md

@@ -6,8 +6,8 @@ In *vars/ondemand-config.yml.example*:
 - adjust `ood_apps`
   - check `version` and `k8s_container`
   - enable required apps
-  - set which images to pre-pull
-- set `enable_pod_prepull` if desired (should default to on probably)
+  - set which images to pre-pull (don't choose too many, we have limited space currently on the worker nodes and pre-pulling will fail if you exhaust it)
+- set `enable_pod_prepull` if desired (should default to on probably, sometimes we have experience really slow pulls, this helps with that)
 - set `control_plane_flavor`, usually to `balanced1.4cpu8ram` for production
 - set `cluster_worker_count` and `worker_flavor` to have enough capacity for the number of users
 

terraform/terraform.tfvars

@@ -1,6 +1,7 @@
 # flavours:
-#   - 4cpu8ram: e07cfee1-43af-4bf6-baac-3bdf7c1b88f8
-#   - 8cpu16ram: 2d02e6a4-3937-4ed3-951a-8e27867ff53e
+#   - 4cpu8ram: e07cfee1-43af-4bf6-baac-3bdf7c1b88f8 (usually used for testing)
+#   - 8cpu16ram: 2d02e6a4-3937-4ed3-951a-8e27867ff53e (usually used for
+#                production, should be good for up to around 40-50 users)
 #   - 16cpu32ram: 674fa81a-69c7-4bf7-b3a9-59989fb63618
 
 services_flavor_id = "e07cfee1-43af-4bf6-baac-3bdf7c1b88f8"

vars/ondemand-config.yml.example

@@ -1,4 +1,99 @@
 ---
+# paramaters you are most likely to want to change are in this top section
+
+# number of training users
+num_users_create: 6
+# number of trainer users
+num_trainers_create: 3
+
+# control plane flavour, usually set to
+# - "balanced1.2cpu4ram" for testing
+# - "balanced1.4cpu8ram" for production
+control_plane_flavor: balanced1.2cpu4ram
+
+# You must have enough worker resources to accommodate the number of users.
+# Currently all apps are limited to 2cpu4ram (although this may change).
+# You cannot allocate the full ram on a worker node, for example we could fit
+# 15x 2cpu4ram sessions on a balanced1.32cpu64ram node. It is recommended to
+# use a small number of large worker nodes to avoid repeated wasting of the
+# overhead, e.g.
+# - 2x balanced1.32cpu64ram for up to 30 2cpu4ram sessions
+# - 3x balanced1.32cpu64ram for up to 45 2cpu4ram sessions
+cluster_worker_count: 1
+worker_flavour: balanced1.8cpu16ram
+
+# apps config
+# - `k8s_container` should be the container that will be used by the app, usually
+#   defined in setup.yml in the git repo
+# - `repo` is the git repo for the app, use the HTTPS address not the SSH one
+# - `version` is the git tag (or commit hash) for the git `repo`
+# - `enabled` apps will show up in open ondemand
+# - `pre_pull` will pull (and cache) the `k8s_container` image on all worker
+#   nodes, so starting those apps will be much faster (`enable_pod_pre_pull`
+#   must also be true)
+#
+# NOTE: at this time the k8s nodes only have 30GB disk space, do not set all
+#       images to pre_pull otherwise the pre-puller will fail, just pick the
+#       one you are actually going to use
+#
+# NOTE2: it is recommended to leave the containers app disabled unless you are
+#        using it, since it only works properly in privileged mode (which is
+#        recommended to be turned off unless using the containers app)
+ood_apps:
+  jupyter_ml101:
+    k8s_container: ghcr.io/nesi/training-environment-jupyter-ml101-app:v0.2.2
+    repo: https://github.com/nesi/training-environment-jupyter-ml101-app.git
+    version: 'v0.2.2'
+    enabled: true
+    pre_pull: false
+
+  jupyter_ml102:
+    k8s_container: ghcr.io/nesi/training-environment-jupyter-ml102-app:v0.2.0
+    repo: https://github.com/nesi/training-environment-jupyter-ml102-app.git
+    version: 'v0.2.0'
+    enabled: true
+    pre_pull: false
+
+  rstudio_rnaseq:
+    k8s_container: ghcr.io/nesi/training-environment-rstudio-rnaseq-app:v0.2.2
+    repo: https://github.com/nesi/training-environment-rstudio-rnaseq-app.git
+    version: 'v0.2.2'
+    enabled: true
+    pre_pull: false
+
+  rstudio:
+    k8s_container: ghcr.io/nesi/training-environment-rstudio-app:v0.3.0
+    repo: https://github.com/nesi/training-environment-rstudio-app.git
+    version: 'v0.3.0'
+    enabled: true
+    pre_pull: false
+
+  shell4b:
+    k8s_container: ghcr.io/nesi/training-environment-jupyter-intermediate-shell-app:v0.3.3
+    repo: https://github.com/nesi/training-environment-jupyter-intermediate-shell-app.git
+    version: 'v0.3.3'
+    enabled: true
+    pre_pull: false
+
+  containers:
+    k8s_container: ghcr.io/nesi/training-environment-jupyter-containers-app:v0.1.0
+    repo: https://github.com/nesi/training-environment-jupyter-containers-app.git
+    version: 'v0.1.0'
+    enabled: false
+    pre_pull: false
+
+# this is currently required for containers app to run properly (fakeroot)
+# Note: you should probably set to false unless you are running a containers workshop
+enable_privileged_pods: false
+
+# pull the images defined in ood_apps onto all k8s worker nodes
+# Note: make sure the worker nodes have enough disk space (especially if many apps are enabled)
+enable_pod_pre_pull: false
+
+###############################################################################
+# variables below this line should not usually need to be changed
+###############################################################################
+
 keycloak_admin_password: "CHANGEME_KEYCLOAK_ADMIN_PASSWORD"
 ldap_admin_password: "CHANGEME_LDAP_ADMIN_PASSWORD"
 
@@ -10,10 +105,6 @@ clouds_yaml_local_location: ~/.config/openstack/clouds.yaml
 
 cluster_rdc_project: NeSI-Training-Test
 
-# number of users
-num_users_create: 6
-num_trainers_create: 3
-
 # authentication settings
 oidc_uri: "/oidc"
 oidc_settings_samefile: true
@@ -77,58 +168,6 @@ clusters:
         strict_host_checking: false
         tmux_bin: /usr/bin/tmux
 
-# apps config
-ood_apps:
-  jupyter_ml101:
-    k8s_container: ghcr.io/nesi/training-environment-jupyter-ml101-app:v0.2.2
-    repo: https://github.com/nesi/training-environment-jupyter-ml101-app.git
-    version: 'v0.2.2'
-    enabled: true
-    pre_pull: false
-
-  jupyter_ml102:
-    k8s_container: ghcr.io/nesi/training-environment-jupyter-ml102-app:v0.2.0
-    repo: https://github.com/nesi/training-environment-jupyter-ml102-app.git
-    version: 'v0.2.0'
-    enabled: true
-    pre_pull: false
-
-  rstudio_rnaseq:
-    k8s_container: ghcr.io/nesi/training-environment-rstudio-rnaseq-app:v0.2.2
-    repo: https://github.com/nesi/training-environment-rstudio-rnaseq-app.git
-    version: 'v0.2.2'
-    enabled: true
-    pre_pull: false
-
-  rstudio:
-    k8s_container: ghcr.io/nesi/training-environment-rstudio-app:v0.3.0
-    repo: https://github.com/nesi/training-environment-rstudio-app.git
-    version: 'v0.3.0'
-    enabled: true
-    pre_pull: false
-
-  shell4b:
-    k8s_container: ghcr.io/nesi/training-environment-jupyter-intermediate-shell-app:v0.3.3
-    repo: https://github.com/nesi/training-environment-jupyter-intermediate-shell-app.git
-    version: 'v0.3.3'
-    enabled: true
-    pre_pull: false
-
-  containers:
-    k8s_container: ghcr.io/nesi/training-environment-jupyter-containers-app:v0.1.0
-    repo: https://github.com/nesi/training-environment-jupyter-containers-app.git
-    version: 'v0.1.0'
-    enabled: false
-    pre_pull: false
-
-# this is currently required for containers app to run properly (fakeroot)
-# Note: you should probably set to false unless you are running a containers workshop
-enable_privileged_pods: false
-
-# pull the images defined in ood_apps onto all k8s worker nodes
-# Note: make sure the worker nodes have enough disk space (especially if many apps are enabled)
-enable_pod_pre_pull: false
-
 # HTTPS LetsEncrypt Settings
 
 acme_directory: https://acme-v02.api.letsencrypt.org/directory