Tero Marttila

e2e kontena service shutdown spec sometimes results in client errors

1

The graceful shutdown during the deploy sometimes causes a client to get an EOF for some request. It is unknown what causes this: it's either a bug in the e2e...

Duplicate/ghost containers in kontena container list

3

It is possible to end up with duplicate containers visible in `kontena container list`. This happened randomly while running e2e tests, and creating/removing the same stack. ``` CONTAINER_ID IMAGE COMMAND...

The first `initial_size` nodes that join the grid are given low `node_number` `1..initial_size` and become initial members. The etcd cluster is formed by each such initial node launching `kontena-etcd` with...

Use per-node tokens for agent-server websocket authentication

4

The current agent-server websocket RPC mechanism uses a grid-level shared secret token for authentication, together with the agent-provided Docker ID for identifying nodes. This is useful for simplifying autoscaling deployments,...

CLI: enhance kontena stack deploy to report service deploy progress, instance deploy errors

11

Using the new `service_deploys -> instance_deploys` given by the server API.

For a stateful service bind-mounting a file that exists in the container image: ``` stack: test/stateful-bind-file services: test: image: redis stateful: true volumes: - /tmp/bashrc:/root/.bashrc ``` This fails to create...

Kernel panic on CoreOS stable 1668.4.0, beta 1688.3.0 with Weave

1

The CoreOS beta 1688.3.0 Linux 4.14.24-coreos kernel will panic as soon as two weave nodes establish an active vxlan connection. This is a CoreOS issue that is not specific to...

The `kontena/lb` image will only enable the http listener if linked to any http services, configured with an `KONTENA_LB_HEALTH_URI`, or if configured with `SSL_CERTS`/`SSL_CERT_*`: https://github.com/kontena/kontena-loadbalancer/blob/8c38532ddadb71e0d92f0debd3a77eb9ece36994/lib/kontena/templates/haproxy/main.text.erb#L32-L34 Where those SSL certs were...

Agent etcd will not be able to recover if the kontena-etcd container was removed while keeping the kontena-etcd-data container

2

The `Kontena::Launchers::Etcd` launcher uses [`#update_membership`](https://github.com/kontena/kontena/blob/master/agent/lib/kontena/launchers/etcd.rb#L111) to determine the etcd `--initial-cluster-state=new|existing` used when creating a new `kontena-etcd` container, and also manage the node's etcd cluster membership if the etcd cluster is...

Use service deploys to to have the CLI service start/stop commands wait for the service containers to start/stop

4

Requires #3299 Fix service scale to set service running state Maybe fixes #2710 to cancel running service deploys if service is redeployed Change the `kontena service start/stop` => `POST /v1/services/.../{start,stop}`=>...