v0.1.0-rc2

We are happy to announce the second release candidate for kube-router 0.1.0

In this release we have focused on introducing instrumentation to monitor metrics and health in kube-router for ease of operations, some minor bug fixes, adjustments to the logging output as well as introduced the BGP graceful restart function to enable zero downtime upgrades in supported environments.

Action required

If you previously used the metrics on port 8080 and wish to keep using this port you must set it at kube-router start using --metrics-port as the default now is 0, disabled

Logging

A lot of the output from kube-router has had logging levels added to prevent spamming the logs with normal operations and debug output. To run kube-router in verbose mode for debugging please add --v=3 to the start options.

Metrics

We have added a sample Grafana dashboard & the possibility to change what port and path the metrics are exposed at. You can read more about the functionality here

Health checks

kube-router now comes with togglable health endpoint to monitor the status of your kube-router deployment. Read more about the functionality and how to enable it here

Local services

Support for service.spec externalTrafficPolicy=Local has been added. It takes precedence over kube-router.io/service.local annotation. Read more about externalTrafficPolicy here

BGP graceful restart

Support has been added in this version for graceful BGP restarts. To enable it start kube-router with --bgp-graceful-restart. When peering with global peers make sure they support this mode before enabling

Notable Bug Fixes

• A regression was introduced in RC1 where GoBGP would panic when adding nodes in some cases. This is fixed in this version by #296
• Conntrack cleanup would log errors when cleaning up old conntrack recods if 0 flows was present, fixed by #305

Changelog

a466ace Nsc conntrack fix (#305)
e25c174 support service.spec externalTrafficPolicy=Local. Takes precedence over kube-router.io/service.local annotation. (#303)
617c773 code cleanup (#301)
3369890 reverted rollingUpdate strategy (#300)
a480a51 Merge pull request #298 from roffe/healthcheck
daefc81 updated help text
b1f3404 fixed bool values for if http health is enabled
38f45a4 chnged max unavail to 1 according to #171
deba3fa mutex on stats
311d726 cleanup health checks daemonsets
bb49531 liveness checks
2c3510d Docs
23f4313 Updated daemonsets to include health checks
258cf2c health docs
f38dc79 Docs for healthz
f2da445 added startup delay before healthchecks starts
cf7c66e Adde flag so health controller knows if HTTP is requested or not
7066d4d change so no http listener starts if port is not in valid range
e53aef2 more work on healthchecks
77cb340 missing bracket
4f8f83d metrics controller ticker
12aec99 added heartbeats
3eb5461 adding output messages
3469cc1 fixed healthport var
e6b01d5 added health controller
f1cb675 Update MAINTAINER.md
193776c prevent calling gobgp AddNeighbour call before GoBGP server is properly started (#296)
f3e7ace Metrics + Logging update (#294)
1492f0b small update to the metrics documentation
60f67d8 Metrics updates (#292)
edda2b1 Added more metrics and small bugfix + more docs (#291)
e36759e Documentation updates (#289)
c8c19e4 Added cli config for changing prometheus port & path (#288)
42f7177 Updated generic cluster docs & daemonset (#284)
8fe0653 Small docs fixes for setting up a generic cluster (#283)
1edee93 Documentation how to install kube-router on generic clusters (#281)
e51dfbf support node local service without SNAT, so the pod can see client origin IP if the request from external IP (#277)
aeaa703 Fix typo in document (#276)
78588fc tunnel interface names cannot be longer than 15 characters (#274)
5155c8d Update README.md
862b410 Make kubeadm install docs clearer (#270)
c07df49 unit tests for syncInternalPeers() (#268)
0b739e4 project status update (#269)

---

Automated with GoReleaser
Built with go version go1.9.3 linux/amd64