From 226f8334d0b5a82b03b2f3a7fe6422b3ddf5535a Mon Sep 17 00:00:00 2001 From: Jennifer Travinski Date: Fri, 11 Mar 2022 17:50:01 -0500 Subject: [PATCH 1/9] Created new section for RKE1 v. RKE2 difference highlights --- .../_index.md | 28 +++++++++++++++++++ 1 file changed, 28 insertions(+) create mode 100644 content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md diff --git a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md new file mode 100644 index 00000000000..28db222307d --- /dev/null +++ b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md @@ -0,0 +1,28 @@ +--- +title: Behavior Differences Between RKE1 and RKE2 +weight: 2450 +--- + +RKE2, also known as RKE Government, is a Kubernetes distribution that focuses on security and compliance for U.S. Federal Government entities. It is considered the next iteration of the Rancher Kubernetes Engine, now known as RKE1. + +RKE1 and RKE2 have several slight behavioral differences to note, and this page will highlight some of these at a high level. + +### Control Plane Components + +RKE1 uses Docker for deploying and managing control plane components and the container runtime for Kubernetes. By contrast, RKE2 launches control plane components such as static pods that are managed by the kubelet. RKE2's container runtime is containerd, which allows things such as container registry mirroring (RKE1 with Docker does not). + +### Cluster API + +RKE2/K3s V2 provisioning is built on top of the Cluster API (CAPI) upstream framework which often makes RKE2-provisioned clusters behave differently than RKE1-provisioned clusters. + +When you make changes to your cluster configuration in RKE2, this may result in nodes reprovisioning. This is controlled by CAPI controllers and not by Rancher itself. Note that for etcd nodes, the same behavior does not apply. + +Users who are used to RKE1 provisioning should take note of this new RKE2 behavior which may be unexpected. + +### Terminology + +You will notice that some terms have changed or gone away going from RKE1 to RKE2. For example, in RKE2, **node templates** have been deprecated, and you can confiugure your cluster nodes directly. Another example is that the term **node pool** in RKE1 is now known as **machine pool** in RKE2. + + + + From e155f2531b1885cbdf20b054ae3cb68fa7f87984 Mon Sep 17 00:00:00 2001 From: Jennifer Travinski Date: Fri, 11 Mar 2022 18:22:57 -0500 Subject: [PATCH 2/9] Updated per feedback --- .../behavior-differences-between-rke1-and-rke2/_index.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md index 28db222307d..e530f5abc17 100644 --- a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md +++ b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md @@ -13,7 +13,7 @@ RKE1 uses Docker for deploying and managing control plane components and the con ### Cluster API -RKE2/K3s V2 provisioning is built on top of the Cluster API (CAPI) upstream framework which often makes RKE2-provisioned clusters behave differently than RKE1-provisioned clusters. +RKE2/K3s provisioning is built on top of the Cluster API (CAPI) upstream framework which often makes RKE2-provisioned clusters behave differently than RKE1-provisioned clusters. When you make changes to your cluster configuration in RKE2, this may result in nodes reprovisioning. This is controlled by CAPI controllers and not by Rancher itself. Note that for etcd nodes, the same behavior does not apply. From 7cf2e83c27089160fcfeb6c0f78ef54cc95b767e Mon Sep 17 00:00:00 2001 From: Jennifer Travinski Date: Fri, 11 Mar 2022 18:25:08 -0500 Subject: [PATCH 3/9] Emphasizing reprovisioning may occur --- .../behavior-differences-between-rke1-and-rke2/_index.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md index e530f5abc17..bc0e1134ec4 100644 --- a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md +++ b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md @@ -15,7 +15,7 @@ RKE1 uses Docker for deploying and managing control plane components and the con RKE2/K3s provisioning is built on top of the Cluster API (CAPI) upstream framework which often makes RKE2-provisioned clusters behave differently than RKE1-provisioned clusters. -When you make changes to your cluster configuration in RKE2, this may result in nodes reprovisioning. This is controlled by CAPI controllers and not by Rancher itself. Note that for etcd nodes, the same behavior does not apply. +When you make changes to your cluster configuration in RKE2, this **may** result in nodes reprovisioning. This is controlled by CAPI controllers and not by Rancher itself. Note that for etcd nodes, the same behavior does not apply. Users who are used to RKE1 provisioning should take note of this new RKE2 behavior which may be unexpected. From 2d252669a392ed444f6d8e9d61123abcaa373571 Mon Sep 17 00:00:00 2001 From: Jennifer Travinski Date: Fri, 11 Mar 2022 18:36:11 -0500 Subject: [PATCH 4/9] Updated per feedback --- .../behavior-differences-between-rke1-and-rke2/_index.md | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md index bc0e1134ec4..94e2c64bfb4 100644 --- a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md +++ b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md @@ -17,6 +17,12 @@ RKE2/K3s provisioning is built on top of the Cluster API (CAPI) upstream framewo When you make changes to your cluster configuration in RKE2, this **may** result in nodes reprovisioning. This is controlled by CAPI controllers and not by Rancher itself. Note that for etcd nodes, the same behavior does not apply. +The following are some specific example configuration changes that may cause the described behavior: + +- When editing the cluster and enabling `drain before delete`, the existing control plane nodes and worker are deleted and new nodes are created. + +- When nodes are being provisioned, performing a scale down operation may result in both nodes getting deleted and a new one being provisioned. + Users who are used to RKE1 provisioning should take note of this new RKE2 behavior which may be unexpected. ### Terminology From a31e4e8eb9e058a097104640a5cb03edaf9f5851 Mon Sep 17 00:00:00 2001 From: Jennifer Travinski Date: Fri, 11 Mar 2022 18:43:27 -0500 Subject: [PATCH 5/9] Updated per feedback --- .../behavior-differences-between-rke1-and-rke2/_index.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md index 94e2c64bfb4..8b739dfed4a 100644 --- a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md +++ b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md @@ -21,7 +21,7 @@ The following are some specific example configuration changes that may cause the - When editing the cluster and enabling `drain before delete`, the existing control plane nodes and worker are deleted and new nodes are created. -- When nodes are being provisioned, performing a scale down operation may result in both nodes getting deleted and a new one being provisioned. +- When nodes are being provisioned, performing a scale down operation may, in some cases, result in both nodes getting deleted and a new one being provisioned. Users who are used to RKE1 provisioning should take note of this new RKE2 behavior which may be unexpected. From a2af488e27427eec7d3aa6c2dddc80dd08b55f12 Mon Sep 17 00:00:00 2001 From: Jennifer Travinski Date: Fri, 11 Mar 2022 19:14:39 -0500 Subject: [PATCH 6/9] Updated note --- .../behavior-differences-between-rke1-and-rke2/_index.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md index 8b739dfed4a..bce8acd8604 100644 --- a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md +++ b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md @@ -21,7 +21,7 @@ The following are some specific example configuration changes that may cause the - When editing the cluster and enabling `drain before delete`, the existing control plane nodes and worker are deleted and new nodes are created. -- When nodes are being provisioned, performing a scale down operation may, in some cases, result in both nodes getting deleted and a new one being provisioned. +- When nodes are being provisioned and a scale down operation is performed, rather than scaling down directly, nodes are deleted and new nodes are provisioned to reach the desired node count. Users who are used to RKE1 provisioning should take note of this new RKE2 behavior which may be unexpected. From f5cb16986ae339e0dd612b872e40f81644850900 Mon Sep 17 00:00:00 2001 From: Jennifer Travinski Date: Mon, 14 Mar 2022 18:15:23 -0400 Subject: [PATCH 7/9] Fixed typo --- .../behavior-differences-between-rke1-and-rke2/_index.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md index bce8acd8604..28c8d0c741c 100644 --- a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md +++ b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md @@ -27,7 +27,7 @@ Users who are used to RKE1 provisioning should take note of this new RKE2 behavi ### Terminology -You will notice that some terms have changed or gone away going from RKE1 to RKE2. For example, in RKE2, **node templates** have been deprecated, and you can confiugure your cluster nodes directly. Another example is that the term **node pool** in RKE1 is now known as **machine pool** in RKE2. +You will notice that some terms have changed or gone away going from RKE1 to RKE2. For example, in RKE2, **node templates** have been deprecated, and you can configure your cluster nodes directly. Another example is that the term **node pool** in RKE1 is now known as **machine pool** in RKE2. From fcfa473c212fc9133f7bdd92b21b2323ed37f53f Mon Sep 17 00:00:00 2001 From: Jennifer Travinski Date: Thu, 17 Mar 2022 18:13:26 -0400 Subject: [PATCH 8/9] Updated per feedback --- .../behavior-differences-between-rke1-and-rke2/_index.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md index 28c8d0c741c..556bdc53bc4 100644 --- a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md +++ b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md @@ -9,7 +9,7 @@ RKE1 and RKE2 have several slight behavioral differences to note, and this page ### Control Plane Components -RKE1 uses Docker for deploying and managing control plane components and the container runtime for Kubernetes. By contrast, RKE2 launches control plane components such as static pods that are managed by the kubelet. RKE2's container runtime is containerd, which allows things such as container registry mirroring (RKE1 with Docker does not). +RKE1 uses Docker for deploying and managing control plane components, and it also uses Docker as the container runtime for Kubernetes. By contrast, RKE2 launches control plane components such as static pods that are managed by the kubelet. RKE2's container runtime is containerd, which allows things such as container registry mirroring (RKE1 with Docker does not). ### Cluster API @@ -21,13 +21,13 @@ The following are some specific example configuration changes that may cause the - When editing the cluster and enabling `drain before delete`, the existing control plane nodes and worker are deleted and new nodes are created. -- When nodes are being provisioned and a scale down operation is performed, rather than scaling down directly, nodes are deleted and new nodes are provisioned to reach the desired node count. +- When nodes are being provisioned and a scale down operation is performed, rather than scaling down the desired number of nodes, it is possible that the currently provisioning nodes get deleted and new nodes are provisioned to reach the desired node count. Users who are used to RKE1 provisioning should take note of this new RKE2 behavior which may be unexpected. ### Terminology -You will notice that some terms have changed or gone away going from RKE1 to RKE2. For example, in RKE2, **node templates** have been deprecated, and you can configure your cluster nodes directly. Another example is that the term **node pool** in RKE1 is now known as **machine pool** in RKE2. +You will notice that some terms have changed or gone away going from RKE1 to RKE2. For example, in RKE1 provisioning, you use **node templates**; in RKE2 provisioning, you can configure your cluster nodes directly. Another example is that the term **node pool** in RKE1 is now known as **machine pool** in RKE2. From 6e36f7a6b45e31451c3e828f1bdbc893d42765e0 Mon Sep 17 00:00:00 2001 From: Jennifer Travinski Date: Fri, 18 Mar 2022 17:04:02 -0400 Subject: [PATCH 9/9] Updated per feedback --- .../behavior-differences-between-rke1-and-rke2/_index.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md index 556bdc53bc4..99a3f18f370 100644 --- a/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md +++ b/content/rancher/v2.6/en/cluster-provisioning/rke-clusters/behavior-differences-between-rke1-and-rke2/_index.md @@ -9,7 +9,7 @@ RKE1 and RKE2 have several slight behavioral differences to note, and this page ### Control Plane Components -RKE1 uses Docker for deploying and managing control plane components, and it also uses Docker as the container runtime for Kubernetes. By contrast, RKE2 launches control plane components such as static pods that are managed by the kubelet. RKE2's container runtime is containerd, which allows things such as container registry mirroring (RKE1 with Docker does not). +RKE1 uses Docker for deploying and managing control plane components, and it also uses Docker as the container runtime for Kubernetes. By contrast, RKE2 launches control plane components as static pods that are managed by the kubelet. RKE2's container runtime is containerd, which allows things such as container registry mirroring (RKE1 with Docker does not). ### Cluster API @@ -21,13 +21,13 @@ The following are some specific example configuration changes that may cause the - When editing the cluster and enabling `drain before delete`, the existing control plane nodes and worker are deleted and new nodes are created. -- When nodes are being provisioned and a scale down operation is performed, rather than scaling down the desired number of nodes, it is possible that the currently provisioning nodes get deleted and new nodes are provisioned to reach the desired node count. +- When nodes are being provisioned and a scale down operation is performed, rather than scaling down the desired number of nodes, it is possible that the currently provisioning nodes get deleted and new nodes are provisioned to reach the desired node count. Please note that this is a bug in Cluster API, and it will be fixed in an upcoming release. Once fixed, Rancher will update the documentation. Users who are used to RKE1 provisioning should take note of this new RKE2 behavior which may be unexpected. ### Terminology -You will notice that some terms have changed or gone away going from RKE1 to RKE2. For example, in RKE1 provisioning, you use **node templates**; in RKE2 provisioning, you can configure your cluster nodes directly. Another example is that the term **node pool** in RKE1 is now known as **machine pool** in RKE2. +You will notice that some terms have changed or gone away going from RKE1 to RKE2. For example, in RKE1 provisioning, you use **node templates**; in RKE2 provisioning, you can configure your cluster node pools when creating or editing the cluster. Another example is that the term **node pool** in RKE1 is now known as **machine pool** in RKE2.