1.0.0+8bb01ee9ceb7c8fe20cd6075c57a7faf9c66edbe

v1.0.0+8bb01ee9ceb7c8fe20cd6075c57a7faf9c66edbe

Fix Broken Doc Link

1.0.0+6f50d8a7720efc1b31911e96fbb3e8e7c917e44d

Relax stalebot

I think 30 days turned out to be to aggressive given our small team. We had multiple relevant issues closed before we could address them. K8s uses 90 days as well.

1.0.0+2f1b9a60ec87cde2ebf29b511c05562f32dea42b

Merge pull request #142 from notque/master

Correct Test Flag Name

1.0.0+e3df9babb0c4fd229980700fcf2e2f8f14646237

v1.0.0+e3df9babb0c4fd229980700fcf2e2f8f14646237

silences keystone auth log spam using a log adapter

1.0.0+b14b76d4b0b7819113bf0fb2ca57978ffd6a1afb

v1.0.0+b14b76d4b0b7819113bf0fb2ca57978ffd6a1afb

fixes merge conflict

1.0.0+375b3c07cacd17909162e429d503c28e18a10a43

Kubernetes v1.9.0 Support (#135)

This adds support for Kubernetes v1.9.0. No major surprises so far. The
usual nginx smoke test and exposing via LB works.

The Openstack cloud-provider now only supports Cinder v3. The expected
major pain didn't actually kick in. It just works. TM. Famous last
words.

1.0.0+6d036056829719935387bc444ebd96f46c45c224

Add leveled logging (#129)

* Add leveled logging

I reused the `—v` flag for glog to set the desiered log level. By default we only log messages with no level or level 0.
For the flag processing to work I had to move the logger creation into the cobra commands `Run` functions because otherwise I can’t access the flag without hussle.

I moved the logging setup to logutil.NewLogger() to that we can reuse that function in all binaries.

* Remove unused functions

* Refactor wormhole server

use one flat Options struct for both the controller and the tunnel.
They are not meant to be run independently and share some things (e.g. logger)

Also consolidated the server in a single package because

1.0.0+142d149f7e75896a2746459c6347d916eae8b3d5

Upgrade to Kubernetes 1.8.5 (#132)

* Upgrade to Kubernetes 1.8.5

This commit upgrades the helm chart and ignition template to use 1.8.5.

Currently, this assumes that klusters without version information in the
spec are on 1.7.5. All new klusters must provide a version.

With 1.8.5 the default behaviour of the Openstack LoadBalancer creation
changed. Instead of defaulting to an internal LoadBalancer it now
creates an external LB by default. This requires the external network to
be configured in the cloud-provider config.

Fixes #62

* removes v from versions

* fixes logging in case of error while creating kluster struct

* uses one version for new klusters only

1.0.0+f3c4111fe695f9479c7fc0dfa98f45ec5968f6da

removes glog

This commit removes all glog references in favor of the structured
logging approach using go-kit/log.

There is plenty room for improvement. The intention here is to rid glog
without introducing to many bugs.

`kubernikusctl` is left to use glog as the main intent of logging here
is human consumption on the cli without an intermediate log retrieval
system. Also, there's tight integration with k8s clients that log using
glog anyway.

The log output is not yet completely clean and 100% structured logging,
because of the informer framwork being uncooperative. We could get rid
of that using a higher default log threshold instead of the default
`INFO`. That exercise is left for another time.

This commit fixes #100

1.0.0+857118af135aa46be3c87959ac1add2268eda124

v1.0.0+857118af135aa46be3c87959ac1add2268eda124

fixes panic due to missing logger initialization