Fix get build

first round test cases fix

Add travis badge to README.md

Sorry for the big PR

add host network ip address to cache

Corect filtering of OpenStack LBaaS resources to delete

Update meeting schedule for SIG-UI

Merge https://github.com/qbox/cadvisor/tree/ke-release-v0.28.3.

HandleError include the type of the error object

Get docker logs from windows nodes

Update github user link

make: add condition to use docker run -t flag

Update a few more PSC member references

This introduces the following metrics:

    cronjob_controller_sync_one_wall_time_gauge_seconds
    cronjob_controller_sync_one_wall_time_histogram_seconds
    cronjob_controller_sync_all_wall_time_gauge_seconds
    cronjob_controller_sync_all_wall_time_histogram_seconds

For measuring performance of the sync controller loops.

For cronjob_controller_sync_all_wall_time_histogram_seconds, this tends to range between 10-30 seconds depending on rate limit settings. I chose linear buckets from 1-60
seconds to get enough granularity.

For cronjob_controller_sync_one_wall_time_histogram_seconds, this tends to fluctuate exponentially directly proportional to kubeclient rate limiting, so that's why I chose exponential
buckets from 1 microsecond to 1 second.

Implementation note: rate limit saturation isn't actually measured despite the code living in the codebase (https://github.com/lyft/kubernetes/blob/f678a63e9a17580b35805021fe4112f0f4cd6271/pkg/util/metrics/util.go#L68-L74).  Measuring rate limit saturation would help us determine why the CronJobController is slow. Since that's not available, the next best thing is syncOne wall time.

add support ip masquerade of non-overlay addresses; cleanup

…panic

Additional checks to avoid panic due to a non-existant CRD.

…eadme

Add README.md for kernel monitor

Run hack/update-codegen.sh validation