From 5c384c0cd04e7c5c55f4baedfb26a5695d476bcf Mon Sep 17 00:00:00 2001 From: Casey Davenport Date: Mon, 2 Nov 2026 10:24:18 -0500 Subject: [PATCH 1/2] CORE-13923: Fix BIRD probe edits for VXLAN-only manifests (next) --- .../self-managed-onprem/config-options.mdx | 35 ++++++++++++++----- 1 file changed, 27 insertions(+), 8 deletions(-) diff --git a/calico/getting-started/kubernetes/self-managed-onprem/config-options.mdx b/calico/getting-started/kubernetes/self-managed-onprem/config-options.mdx index 098c702b00..3a88a19b0d 100644 --- a/calico/getting-started/kubernetes/self-managed-onprem/config-options.mdx +++ b/calico/getting-started/kubernetes/self-managed-onprem/config-options.mdx @@ -129,24 +129,43 @@ be undone): - Optionally, (to save some resources if you're running a VXLAN-only cluster) completely disable Calico's BGP-based networking: - Replace `calico_backend: "bird"` with `calico_backend: "vxlan"`. This disables BIRD. - - Comment out the line `- -bird-ready` and `- -bird-live` from the calico/node readiness/liveness check (otherwise disabling BIRD will cause the - readiness/liveness check to fail on every node): + - Remove the `- --bird-live` line from the $[nodecontainer] liveness probe and the `- --bird-ready` line from its + readiness probe. Otherwise, disabling BIRD causes both checks to fail on every node, and the kubelet restarts + $[nodecontainer] repeatedly. + +After these edits, the $[nodecontainer] probes look like this: ```yaml livenessProbe: exec: command: - - /bin/calico-node - - -felix-live - - -bird-live + - calico + - component + - node + - health + - --felix-live readinessProbe: exec: command: - - /bin/calico-node - - -bird-ready - - -felix-ready + - calico + - component + - node + - health + - --felix-ready ``` +Manifests from earlier releases ran `/bin/calico-node` with single-dash flags (`-bird-live` and `-bird-ready`). If you +script these edits, update any pattern that matches the old form. + +:::tip + +For a VXLAN-only cluster with BGP disabled, you can skip these edits and start from +[calico-vxlan.yaml]($[manifestsUrl]/manifests/calico-vxlan.yaml) instead. It already sets `calico_backend: "vxlan"` +and leaves the BIRD checks out of the probes. It creates the default IP pool with `CALICO_IPV4POOL_VXLAN` set to +`CrossSubnet`; change the value to `Always` if you need VXLAN for all traffic between nodes. + +::: + For more information on $[nodecontainer]'s configuration variables, including additional VXLAN settings, see [Configuring $[nodecontainer]](../../../reference/configure-calico-node.mdx). From a6583107dbc6a836475cea954ca7863ed4862ef0 Mon Sep 17 00:00:00 2001 From: Casey Davenport Date: Mon, 2 Nov 2026 10:24:24 -0500 Subject: [PATCH 2/2] CORE-13923: Fix BIRD probe edits for VXLAN-only manifests (3.33) --- .../self-managed-onprem/config-options.mdx | 35 ++++++++++++++----- 1 file changed, 27 insertions(+), 8 deletions(-) diff --git a/calico_versioned_docs/version-3.33/getting-started/kubernetes/self-managed-onprem/config-options.mdx b/calico_versioned_docs/version-3.33/getting-started/kubernetes/self-managed-onprem/config-options.mdx index 098c702b00..3a88a19b0d 100644 --- a/calico_versioned_docs/version-3.33/getting-started/kubernetes/self-managed-onprem/config-options.mdx +++ b/calico_versioned_docs/version-3.33/getting-started/kubernetes/self-managed-onprem/config-options.mdx @@ -129,24 +129,43 @@ be undone): - Optionally, (to save some resources if you're running a VXLAN-only cluster) completely disable Calico's BGP-based networking: - Replace `calico_backend: "bird"` with `calico_backend: "vxlan"`. This disables BIRD. - - Comment out the line `- -bird-ready` and `- -bird-live` from the calico/node readiness/liveness check (otherwise disabling BIRD will cause the - readiness/liveness check to fail on every node): + - Remove the `- --bird-live` line from the $[nodecontainer] liveness probe and the `- --bird-ready` line from its + readiness probe. Otherwise, disabling BIRD causes both checks to fail on every node, and the kubelet restarts + $[nodecontainer] repeatedly. + +After these edits, the $[nodecontainer] probes look like this: ```yaml livenessProbe: exec: command: - - /bin/calico-node - - -felix-live - - -bird-live + - calico + - component + - node + - health + - --felix-live readinessProbe: exec: command: - - /bin/calico-node - - -bird-ready - - -felix-ready + - calico + - component + - node + - health + - --felix-ready ``` +Manifests from earlier releases ran `/bin/calico-node` with single-dash flags (`-bird-live` and `-bird-ready`). If you +script these edits, update any pattern that matches the old form. + +:::tip + +For a VXLAN-only cluster with BGP disabled, you can skip these edits and start from +[calico-vxlan.yaml]($[manifestsUrl]/manifests/calico-vxlan.yaml) instead. It already sets `calico_backend: "vxlan"` +and leaves the BIRD checks out of the probes. It creates the default IP pool with `CALICO_IPV4POOL_VXLAN` set to +`CrossSubnet`; change the value to `Always` if you need VXLAN for all traffic between nodes. + +::: + For more information on $[nodecontainer]'s configuration variables, including additional VXLAN settings, see [Configuring $[nodecontainer]](../../../reference/configure-calico-node.mdx).