Skip to content

Commit 319cc9b

Browse files
openshift-merge-bot[bot]openshift-merge-bot[bot]
authored
Merge pull request #1199 from eduolivares/bgp-adoption-docs
Add BGP-specific networking documentation for adoption
2 parents b374ad7 + 61a7e27 commit 319cc9b

4 files changed

+110
-1
lines changed

‎docs_user/modules/proc_adopting-compute-services-to-the-data-plane.adoc‎

Lines changed: 18 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -774,6 +774,24 @@ node is correctly listed in the `OpenStackDataPlaneNodeSet` CR.
774774
to use the same profile as the one set on the {OpenStackShort} {rhos_prev_ver} node.
775775
====
776776

777+
+
778+
[IMPORTANT]
779+
====
780+
For environments that are enabled with border gateway protocol (BGP), preserve the default routes on the data plane nodes.
781+
782+
When adopting {rhos_prev_long} {rhos_prev_ver} environments with BGP, default routes can be lost when the data plane adoption procedure stops the {rhos_prev_long} services, specifically when FRRouting (FRR) is stopped. This causes connectivity issues during the {rhos_acro} data plane deployment.
783+
784+
To prevent this, configure the required routes by using `os-net-config` on the data plane nodes (Compute nodes and Networker nodes) affected by this issue.
785+
786+
Modify your `os-net-config` configuration file by adding the required routes, and then apply it:
787+
788+
----
789+
$ sudo os-net-config -c /etc/os-net-config/modified_config_with_routes.yaml --provider ifcfg
790+
----
791+
792+
This temporary default route is needed during the installation of the first services (such as `download-cache`) and is removed when the `configure-network` service applies the new network configuration.
793+
====
794+
777795
. Remove the remaining {OpenStackPreviousInstaller} services:
778796

779797
.. Create an `OpenStackDataPlaneService` CR to clean up the data plane services you are adopting:

‎docs_user/modules/proc_adopting-networker-services-to-the-data-plane.adoc‎

Lines changed: 17 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -259,6 +259,23 @@ EOF
259259
* `edpm_ovn_bridge_mappings: <bridge_mappings>` specifies the bridge mapping values that you used in your {rhos_prev_long} {rhos_prev_ver} deployment.
260260
* `edpm_enable_chassis_gw` specifies whether to run `ovn-controller` in gateway mode.
261261
+
262+
[IMPORTANT]
263+
====
264+
For environments that are enabled with border gateway protocol (BGP), preserve the default routes on the data plane nodes.
265+
266+
When adopting {rhos_prev_long} {rhos_prev_ver} environments with BGP, default routes can be lost when the data plane adoption procedure stops the {rhos_prev_long} services, specifically when FRRouting (FRR) is stopped. This causes connectivity issues during the {rhos_acro} data plane deployment.
267+
268+
To prevent this, configure the required routes by using `os-net-config` on the data plane nodes (Compute nodes and Networker nodes) affected by this issue.
269+
270+
Modify your `os-net-config` configuration file by adding the required routes, and then apply it:
271+
272+
----
273+
$ sudo os-net-config -c /etc/os-net-config/modified_config_with_routes.yaml --provider ifcfg
274+
----
275+
276+
This temporary default route is needed during the installation of the first services (such as `download-cache`) and is removed when the `configure-network` service applies the new network configuration.
277+
====
278+
+
262279
[NOTE]
263280
====
264281
For environments that are enabled with border gateway protocol (BGP), you must add the following services to the `services` list in the order shown:

‎docs_user/modules/proc_configuring-networking-for-control-plane-services.adoc‎

Lines changed: 33 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -97,4 +97,36 @@ balancer IP addresses.
9797
The example above would exclude the `172.17.0.65` address from the allocation
9898
pool.
9999

100-
// TODO: is there anything specific to mention about BGP L3 mode here?
100+
. For environments that are enabled with border gateway protocol (BGP), add routes to the `NetworkAttachmentDefinition` CRs so that the pods can communicate with the {rhos_prev_long} Controller nodes and Compute nodes over the isolated networks. This is similar to the routes that should be added to the `NodeNetworkConfigurationPolicy` CRs in BGP environments. For more information about isolated networks, see xref:configuring-openshift-worker-nodes_{context}[Configuring isolated networks on RHOCP worker nodes]. The following example shows a `NetworkAttachmentDefinition` CR for the storage network with routes:
101+
+
102+
----
103+
apiVersion: k8s.cni.cncf.io/v1
104+
kind: NetworkAttachmentDefinition
105+
metadata:
106+
name: storage
107+
namespace: openstack
108+
spec:
109+
config: |
110+
{
111+
"cniVersion": "0.3.1",
112+
"name": "storage",
113+
"type": "bridge",
114+
"isDefaultGateway": false,
115+
"isGateway": true,
116+
"forceAddress": false,
117+
"hairpinMode": true,
118+
"ipMasq": false,
119+
"bridge": "storage",
120+
"ipam": {
121+
"type": "whereabouts",
122+
"range": "172.18.0.0/24",
123+
"range_start": "172.18.0.30",
124+
"range_end": "172.18.0.70",
125+
"routes": [
126+
{"dst": "172.31.0.0/24", "gw": "172.18.0.1"},
127+
{"dst": "192.168.188.0/24", "gw": "172.18.0.1"},
128+
{"dst": "99.99.0.0/16", "gw": "172.18.0.1"}
129+
]
130+
}
131+
}
132+
----

‎docs_user/modules/proc_configuring-openshift-worker-nodes.adoc‎

Lines changed: 42 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -87,3 +87,45 @@ items:
8787
kubernetes.io/hostname: ocp-worker-0
8888
node-role.kubernetes.io/worker: ""
8989
----
90+
+
91+
[NOTE]
92+
====
93+
For environments that are enabled with border gateway protocol (BGP), you might need to add additional routes in the `NodeNetworkConfigurationPolicy` CR so that {OpenShiftShort} worker nodes can reach the {rhos_prev_long} Controller nodes and Compute nodes over the control plane and internal API networks.
94+
95+
When you configure the {OpenShiftShort} worker nodes network in the `NodeNetworkConfigurationPolicy` CR, add routes for each of the following networks:
96+
97+
* External network (for example, `172.31.0.0/24`)
98+
* Control plane network (for example, `192.168.188.0/24`)
99+
* BGP main network (for example, `99.99.0.0/16`)
100+
101+
The following example shows the `routes.config` section from a `NodeNetworkConfigurationPolicy` CR for a worker node with BGP configured. In this example, `100.64.0.17` and `100.65.0.17` are the IP addresses of the leaf switches that are connected to the specific {OpenShiftShort} node:
102+
103+
----
104+
routes:
105+
config:
106+
- destination: 99.99.0.0/16
107+
next-hop-address: 100.64.0.17
108+
next-hop-interface: enp7s0
109+
weight: 200
110+
- destination: 99.99.0.0/16
111+
next-hop-address: 100.65.0.17
112+
next-hop-interface: enp8s0
113+
weight: 200
114+
- destination: 172.31.0.0/24
115+
next-hop-address: 100.64.0.17
116+
next-hop-interface: enp7s0
117+
weight: 200
118+
- destination: 172.31.0.0/24
119+
next-hop-address: 100.65.0.17
120+
next-hop-interface: enp8s0
121+
weight: 200
122+
- destination: 192.168.188.0/24
123+
next-hop-address: 100.64.0.17
124+
next-hop-interface: enp7s0
125+
weight: 200
126+
- destination: 192.168.188.0/24
127+
next-hop-address: 100.65.0.17
128+
next-hop-interface: enp8s0
129+
weight: 200
130+
----
131+
====

0 commit comments

Comments
 (0)