Current docs still state that Calico uses etcd as the backend, makes clear that this only holds true for versions below 1.12 and adds a stanza around using APIServer without Typha for 1.12 onwards.
I'm hoping this will cut down on the confusion when reading the master branch's docs and seeing features not included in the latest stable Kops release..
A stopgap until we can properly version our documentation.
As per Lance Robson's message in Slack,
https://kubernetes.slack.com/archives/C3QUFP0QM/p1537959279000100?thread_ts=1537951563.000100&cid=C3QUFP0QM
above 50-100 nodes Calico recommends usage of BGP route reflectors:
> Clearly I'm not impartial here, but I can tell you that we test Calico to 5K
> nodes.
> We recommend BGP route reflectors once you go above ~50-100 nodes (we test up
> to 100 nodes without RRs). The reason for RRs is that by default calico sets up
> BGP connections in a full mesh - clearly the number of connections will scale as
> the number of nodes squared. With RRs, the number of BGP connections scales
> linearly with nodes - but then you have to manage the RRs too.
> The next version of Calico (v3.3 - code is in master already, but not heavily
> tested yet) will bring a feature to make RRs much easier to deploy - just
> annotate a few nodes and they become RRs.
Adding this detail to documentation as I recently got a cluster with such
issues.
fixes#2606
Most part of the changes are similar to current supported CNI networking
provider. Kube-router also support IPVS bassed service proxy which can
be used as replacement for kube-proxy. So the manifest for kube-router
included with this patch enables kube-router to provide pod-to-pod
networking, IPVS based service proxy and ingress pod firewall.
This gives the ability to a user to enable cross-subnet mode in Calico.
Also introduces a new addon that, full disclosure, I wrote.
[ottoyiu/k8s-ec2-srcdst](https://github.com/ottoyiu/k8s-ec2-srcdst)
* Integrating Canal (Flannel + Calico) for CNI
Initial steps to integrate Canal as a CNI provider for kops
Removed CNI in help as per chrislovecnm
* Integration tests, getting closer to working
- Added some integration tests for Canal
- Finding more places Canal needed to be added
- Sneaking in update to Calico Policy Controller
* Add updated conversion file
* turned back on canal integration tests
* fixed some rebase issues
* Fixed tests and flannel version
* Fixed canal yaml, and some rebasing errors
- Added some env vars to the install-cni container to get the proper
node name handed off
* Added resource limits
- set resource limits on containers for Canal
- Ran through basic calico tutorials to verify functionality
* Updating Calico parts to Calico 2.0.2
Adding the option to install Calico with the `--networking calico`
argument. This will currently deploy Calico v2.0 to the cluster.
Documentation has also been updated with information about Calico and
where one can find more information or help.
Reilly Brogan
GuyTempleton
Peter Rifel
misdre
Artem Yarmoliuk
Kenjiro Nakayama
k8s-ci-robot
Erik Stidham
Vlad Ionescu
SataQiu
Byungjin Park
sunlintong
fqsghostcloud
Christian Kampka
Maciej Kwiek
Federico Hernandez
Raffaele Di Fazio
AdamDang
Felipe Cavalcanti
Reid Wiggins
Manuel de Brito Fontes
Andy Randall
Caleb Gilmour
Justin Santa Barbara
Murali Reddy
Murali Reddy
Chris Love
Jordan Jennings
Otto Yiu
Shimi Kiviti
Voyalab LLC
Jon Chiu
Matthew Mihok
Stephen Schlie
Sharif Nassar
Jakub Paweł Głazik