Merge branch 'master' into k8s2d

prepare-vms/lib/commands.sh

@@ -53,11 +53,18 @@ _cmd_deploy() {
 
     echo deploying > tags/$TAG/status
     sep "Deploying tag $TAG"
+
+    # Wait for cloudinit to be done
+    pssh "
+    while [ ! -f /var/lib/cloud/instance/boot-finished ]; do
+        sleep 1
+    done"
+
+    # Copy settings and install Python YAML parser
     pssh -I tee /tmp/settings.yaml <tags/$TAG/settings.yaml
     pssh "
     sudo apt-get update &&
-    sudo apt-get install -y python-setuptools &&
-    sudo easy_install pyyaml"
+    sudo apt-get install -y python-yaml"
 
     # Copy postprep.py to the remote machines, and execute it, feeding it the list of IP addresses
     pssh -I tee /tmp/postprep.py <lib/postprep.py
@@ -178,6 +185,14 @@ EOF"
     sep "Done"
 }
 
+_cmd kubereset "Wipe out Kubernetes configuration on all nodes"
+_cmd_kubereset() {
+    TAG=$1
+    need_tag
+
+    pssh "sudo kubeadm reset --force"
+}
+
 _cmd kubetest "Check that all nodes are reporting as Ready"
 _cmd_kubetest() {
     TAG=$1
@@ -261,6 +276,15 @@ _cmd_opensg() {
     infra_opensg
 }
 
+_cmd pssh "Run an arbitrary command on all nodes"
+_cmd_pssh() {
+    TAG=$1
+    need_tag
+    shift
+
+    pssh "$@"
+}
+
 _cmd pull_images "Pre-pull a bunch of Docker images"
 _cmd_pull_images() {
     TAG=$1
@@ -376,6 +400,22 @@ _cmd_test() {
     test_tag
 }
 
+# Sometimes, weave fails to come up on some nodes.
+# Symptom: the pods on a node are unreachable (they don't even ping).
+# Remedy: wipe out Weave state and delete weave pod on that node.
+# Specifically, identify the weave pod that is defective, then:
+# kubectl -n kube-system exec weave-net-XXXXX -c weave rm /weavedb/weave-netdata.db
+# kubectl -n kube-system delete pod weave-net-XXXXX
+_cmd weavetest "Check that weave seems properly setup"
+_cmd_weavetest() {
+    TAG=$1
+    need_tag
+    pssh "
+    kubectl -n kube-system get pods -o name | grep weave | cut -d/ -f2 |
+    xargs -I POD kubectl -n kube-system exec POD -c weave -- \
+    sh -c \"./weave --local status | grep Connections | grep -q ' 1 failed' || ! echo POD \""
+}
+
 greet() {
     IAMUSER=$(aws iam get-user --query 'User.UserName')
     info "Hello! You seem to be UNIX user $USER, and IAM user $IAMUSER."
@@ -501,12 +541,3 @@ make_tag() {
     fi
     date +%Y-%m-%d-%H-%M-$USER
 }
-
-describe_tag() {
-    FIXME
-    # Display instance details and reachability/status information
-    TAG=$1
-    need_tag $TAG
-    aws_display_instances_by_tag $TAG
-    aws_display_instance_statuses_by_tag $TAG
-}

prepare-vms/lib/infra/aws.sh

@@ -94,14 +94,11 @@ wait_until_tag_is_running() {
     done_count=0
     while [[ $done_count -lt $COUNT ]]; do
         let "i += 1"
-        info "$(printf "%d instances online" $done_count)"
+        info "$(printf "%d/%d instances online" $done_count $COUNT)"
         done_count=$(aws ec2 describe-instances \
-            --filters "Name=instance-state-name,Values=running" \
-            "Name=tag:Name,Values=$TAG" \
-            --query "Reservations[*].Instances[*].State.Name" \
-            | tr "\t" "\n" \
-            | wc -l)
-
+            --filters "Name=tag:Name,Values=$TAG" \
+            "Name=instance-state-name,Values=running" \
+            --query "length(Reservations[].Instances[])")
         if [[ $i -gt $max_retry ]]; then
             die "Timed out while waiting for instance creation (after $max_retry retries)"
         fi
@@ -165,7 +162,6 @@ aws_get_instance_ids_by_filter() {
 
 aws_get_instance_ids_by_client_token() {
     TOKEN=$1
-    need_tag $TOKEN
     aws_get_instance_ids_by_filter Name=client-token,Values=$TOKEN
 }
 

prepare-vms/lib/postprep.py

@@ -83,7 +83,7 @@ system("sudo sed -i 's/PasswordAuthentication no/PasswordAuthentication yes/' /e
 
 system("sudo service ssh restart")
 system("sudo apt-get -q update")
-system("sudo apt-get -qy install git jq python-pip")
+system("sudo apt-get -qy install git jq")
 
 #######################
 ### DOCKER INSTALLS ###
@@ -98,7 +98,6 @@ system("sudo apt-get -q update")
 system("sudo apt-get -qy install docker-ce")
 
 ### Install docker-compose
-#system("sudo pip install -U docker-compose=={}".format(COMPOSE_VERSION))
 system("sudo curl -sSL -o /usr/local/bin/docker-compose https://github.com/docker/compose/releases/download/{}/docker-compose-{}-{}".format(COMPOSE_VERSION, platform.system(), platform.machine()))
 system("sudo chmod +x /usr/local/bin/docker-compose")
 system("docker-compose version")

prepare-vms/settings/swarm.yaml

@@ -20,8 +20,8 @@ paper_margin: 0.2in
 engine_version: stable
 
 # These correspond to the version numbers visible on their respective GitHub release pages
-compose_version: 1.21.1
-machine_version: 0.14.0
+compose_version: 1.22.0
+machine_version: 0.15.0
 
 # Password used to connect with the "docker user"
-docker_user_password: training
+docker_user_password: training

slides/k8s/kubectlrun.md

@@ -32,7 +32,8 @@
 
 --
 
-OK, what just happened?
+(Starting with Kubernetes 1.12, we get a message telling us that
+`kubectl run` is deprecated. Let's ignore it for now.)
 
 ---
 
@@ -228,6 +229,44 @@ We could! But the *deployment* would notice it right away, and scale back to the
 
 ---
 
+## What about that deprecation warning?
+
+- As we can see from the previous slide, `kubectl run` can do many things
+
+- The exact type of resource created is not obvious
+
+- To make things more explicit, it is better to use `kubectl create`:
+
+  - `kubectl create deployment` to create a deployment
+
+  - `kubectl create job` to create a job
+
+- Eventually, `kubectl run` will be used only to start one-shot pods
+
+  (see https://github.com/kubernetes/kubernetes/pull/68132)
+
+---
+
+## Various ways of creating resources
+
+- `kubectl run` 
+
+  - easy way to get started
+  - versatile
+
+- `kubectl create <resource>` 
+
+  - explicit, but lacks some features
+  - can't create a CronJob
+  - can't pass command-line arguments to deployments
+
+- `kubectl create -f foo.yaml` or `kubectl apply -f foo.yaml`
+
+  - all features are available
+  - requires writing YAML
+
+---
+
 ## Viewing logs of multiple pods
 
 - When we specify a deployment name, only one single pod's logs are shown

slides/k8s/namespaces.md

@@ -192,3 +192,38 @@ Note: it might take a minute or two for the app to be up and running.
   ```
 
 ]
+
+---
+
+## Switching namespaces more easily
+
+- Defining a new context for each namespace can be cumbersome
+
+- We can also alter the current context with this one-liner:
+
+  ```bash
+  kubectl config set-context --current --namespace=foo
+  ```
+
+- We can also use a little helper tool called `kubens`:
+
+  ```bash
+  # Switch to namespace foo
+  kubens foo
+  # Switch back to the previous namespace
+  kubens -
+  ```
+
+---
+
+##  `kubens` and `kubectx`
+
+- With `kubens`, we can switch quickly between namespaces
+
+- With `kubectx`, we can switch quickly between contexts
+
+- Both tools are simple shell scripts available from https://github.com/ahmetb/kubectx
+
+- On our clusters, they are installed as `kns` and `kctx`
+
+  (for brevity and to avoid completion clashes between `kubectx` and `kubectl`)

slides/swarm-fullday.yml

@@ -55,6 +55,7 @@ chapters:
   - swarm/apiscope.md
 - - swarm/logging.md
   - swarm/metrics.md
+  - swarm/gui.md
   - swarm/stateful.md
   - swarm/extratips.md
   - shared/thankyou.md

slides/swarm/gui.md

@@ -0,0 +1,51 @@
+# GUI's: Web Admin of Swarms and Registry
+
+What about web interfaces to control and manage Swarm?
+
+- [Docker Enterprise](https://www.docker.com/products/docker-enterprise) is Docker Inc's paid offering, which has GUI's.
+
+- [Portainer](https://portainer.io) is a popular open source web GUI for Swarm with node agents.
+
+- [Portus](http://port.us.org) is a SUSE-backed open source web GUI for registry.
+
+- Find lots of other Swarm tools in the [Awesome Docker list](http://awesome-docker.netlify.com).
+
+---
+
+## Lets deploy Portainer
+
+- Yet another stack file
+
+.exercise[
+
+- Make sure we are in the stacks directory:
+  ```bash
+  cd ~/container.training/stacks
+  ```
+
+- Deploy the Portainer stack:
+  ```bash
+  docker stack deploy -c portainer.yml portainer
+  ```
+
+]
+
+---
+
+## View and setup Portainer
+
+- go to `<node ip>:9090`
+
+- You should see the setup UI. Create a 8-digit password.
+
+- Next, tell Portainer how to connect to docker.
+
+- We'll use the agent method (one per node).
+
+  - For connection, choose `Agent`
+
+  - Name: `swarm1`
+
+  - Agent URL: `tasks.agent:9001`
+
+- Let's browse around the interface

slides/swarm/versions.md

@@ -1,8 +1,8 @@
 ## Brand new versions!
 
-- Engine 18.03
-- Compose 1.21
-- Machine 0.14
+- Engine 18.06
+- Compose 1.22
+- Machine 0.15
 
 .exercise[
 

stacks/portainer.yml

@@ -0,0 +1,15 @@
+version: "3.4"
+
+services:
+  web:
+    image: portainer/portainer
+    ports:
+      - "9000:9000"
+  agent:
+    image: portainer/agent
+    volumes:
+      - /var/run/docker.sock:/var/run/docker.sock
+    deploy:
+      mode: global
+    environment:
+      AGENT_CLUSTER_ADDR: tasks.agent