Change Details

The following things need to be discussed and decided on / acted on / around the dev summit: 1. Documentation on building debian packages of kubernetes, including how to maintain our patches on top of upstream for each release. 2. Move tool labs to run from the debian packages rather than current bash script 3. Decide on upgrade cadence for Docker. Upgrades need to be tested somewhere and then rolled out. Upgrading is probably going to be disruptive in some form, since containers will still need to restart after docker upgrade. 4. Decide on upgrade cadence for kubernetes. This is far less problematic than docker, however. 5. Decide on wether the upgrade cadence for docker / kubernetes will be the same for prod / tools, or different. If different, how different? What way is this going to work? 6. Figure out how to unify the puppet code between tools and production kubernetes. 7. Figure out a real test cluster that we can test changes on.

The following things need to be discussed and decided on / acted on: [] T161031 Documentation on building debian packages of kubernetes, including how to maintain our patches on top of upstream for each release. [] Move tool labs to run from the debian packages rather than current bash script [] Decide on upgrade cadence for Docker. Upgrades need to be tested somewhere and then rolled out. Upgrading is probably going to be disruptive in some form, since containers will still need to restart after docker upgrade. [] Decide on upgrade cadence for kubernetes. This is far less problematic than docker, however. [] Decide on wether the upgrade cadence for docker / kubernetes will be the same for prod / tools, or different. If different, how different? What way is this going to work? [] Figure out how to unify the puppet code between tools and production kubernetes. [] Figure out a real test cluster that we can test changes on.

The following things need to be discussed and decided on / acted on / around the dev summit: 1.[] T161031 Documentation on building debian packages of kubernetes, including how to maintain our patches on top of upstream for each release. 2.[] Move tool labs to run from the debian packages rather than current bash script 3.[] Decide on upgrade cadence for Docker. Upgrades need to be tested somewhere and then rolled out. Upgrading is probably going to be disruptive in some form, since containers will still need to restart after docker upgrade. 4.[] Decide on upgrade cadence for kubernetes. This is far less problematic than docker, however. 5.[] Decide on wether the upgrade cadence for docker / kubernetes will be the same for prod / tools, or different. If different, how different? What way is this going to work? 6.[] Figure out how to unify the puppet code between tools and production kubernetes. 7.[] Figure out a real test cluster that we can test changes on.