Building upon the Rolling Update Scenario present in k1s, this folder contains a couple of scripts to facilitate running a nice Rolling Update demo.
Now uses k1s.py. Checkout k1s.py from https://github.com/mjbright/k1spy
Ideally run the scripts using a tmux multi-pane layout as shown below. Alternatively separate windows could be used for each pane.
This is composed of 2 scripts
- ./rollout_demo.sh: Runs through a scenario showing/pausing (most) commands before running them
- an initial deployment
- scale up
- rolling upgrade to v2
- rollout status
- rollout history
- rolling upgrade to v3
- rollout status
- rollout history
- rollout undo
- ./rollout_demo_curl.sh: Loops
- determining the Service IP address
- Making a curl request to the Service
The default image repo has tags 1, 2, 3 which report container hostname, image name:tag, requesting IP where the image "name:tag" text changes colour dependent upon the image.
Thus the effect of a rolling update is clearly seen.
See below for thoughts about possible variations.
+--------------------------------------------------------------------------------+
| $ ./rollout_demo.sh |
| |
| |
| |
+--------------------------------------------------------------------------------+
| $ ./rollout_demo_curl.sh |
| |
| |
| |
+--------------------------+--------------------------+--------------------------+
| $ ./k1s.py /k1s.py po,dep,rs,svc
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
+--------------------------------------------------------------------------------+- Vary Rolling upgrade parameters (MaxSurge/MaxUnavailable)
- Different Upgrade Strategies
- Recreate
- "Traffic Shift" - requires 1 ReplicaSet per version, modify Service label selector to point to "new" or "old" ReplicaSet
- Implement HealthChecks to prevent occasional "connection refused"
- Useful to demonstrate with/without healthchecks