kubespray/docs/cri-o.md

84 lines
2.2 KiB
Markdown

# CRI-O
[CRI-O] is a lightweight container runtime for Kubernetes.
Kubespray supports basic functionality for using CRI-O as the default container runtime in a cluster.
* Kubernetes supports CRI-O on v1.11.1 or later.
* etcd: configure either kubeadm managed etcd or host deployment
_To use the CRI-O container runtime set the following variables:_
## all/all.yml
```yaml
download_container: false
skip_downloads: false
etcd_kubeadm_enabled: true
```
## k8s_cluster/k8s_cluster.yml
```yaml
container_manager: crio
```
## etcd.yml
```yaml
etcd_deployment_type: host # optionally and mutually exclusive with etcd_kubeadm_enabled
```
## all/crio.yml
Enable docker hub registry mirrors
```yaml
crio_registries_mirrors:
- prefix: docker.io
insecure: false
blocked: false
location: registry-1.docker.io
mirrors:
- location: 192.168.100.100:5000
insecure: true
- location: mirror.gcr.io
insecure: false
```
## Note about pids_limit
For heavily mult-threaded workloads like databases, the default of 1024 for pids-limit is too low.
This parameter controls not just the number of processes but also the amount of threads
(since a thread is technically a process with shared memory). See [cri-o#1921]
In order to increase the default `pids_limit` for cri-o based deployments you need to set the `crio_pids_limit`
for your `k8s_cluster` ansible group or per node depending on the use case.
```yaml
crio_pids_limit: 4096
```
[CRI-O]: https://cri-o.io/
[cri-o#1921]: https://github.com/cri-o/cri-o/issues/1921
## Note about user namespaces
CRI-O has support for user namespaces. This feature is optional and can be enabled by setting the following two variables.
```yaml
crio_runtimes:
- name: runc
path: /usr/bin/runc
type: oci
root: /run/runc
allowed_annotations:
- "io.kubernetes.cri-o.userns-mode"
crio_remap_enable: true
```
The `allowed_annotations` configures `crio.conf` accordingly.
The `crio_remap_enable` configures the `/etc/subuid` and `/etc/subgid` files to add an entry for the **containers** user.
By default, 16M uids and gids are reserved for user namespaces (256 pods * 65536 uids/gids) at the end of the uid/gid space.