Upgrading EDA#

Assuming you have a working EDA cluster the upgrade procedure will consist of the following steps:

1. Pause NPP interactions.
2. Backup your existing cluster.
3. Update the playground repository.
4. Uninstall the existing version of EDA.
5. Install the new EDA kpt package (on both active and standby members if running geo redundant).
6. Restore your backup.
7. Upgrade your applications.
8. Resume NPP interactions.

Pausing NPP interactions#

Prior to taking a backup of your cluster, place all TopoNode resources into emulate mode to avoid any ongoing interactions with the network devices during the upgrade process.

In this mode, EDA does not interact with target devices, effectively pausing the cluster's interaction with your infrastructure. You can still interact with EDA and the TopoNode resources; changes are pushed upon switching back to normal mode.

To set emulate mode in bulk, run the script from the playground repo directory on a machine where you have kubectl configured with the access to your cluster:

make set-npp-mode-emulate

After the script has been run, verify that the TopoNode resources are in emulate mode:

kubectl get toponode -A \
-o custom-columns='NAMESPACE:.metadata.namespace,NAME:.metadata.name,MODE:.spec.npp.mode'

NAMESPACE       NAME     MODE
my-other-ns     leaf1    emulate
my-other-ns     leaf2    emulate
my-other-ns     leaf3    emulate
my-other-ns     leaf4    emulate
my-other-ns     spine1   emulate
my-other-ns     spine2   emulate
eda             leaf1    emulate
eda             leaf2    emulate
eda             spine1   emulate

Backing up your cluster#

Backing up your existing cluster is performed using the collect-backup target provided in the Makefile of the playground repository.

make collect-backup

[ INFO ] Starting backup
Platform backup done at /tmp/eda-platform-backup-2025-12-18-21-37-42.tar.gz
[  OK  ] Collected backup
[ INFO ] Transferring to host /tmp/eda-support/logs-2025-12-18/eda-platform-backup-2025-12-18-21-37-42.tar.gz
tar: Removing leading `/' from member names
[  OK  ] Transferred to /tmp/eda-support/logs-2025-12-18/eda-platform-backup-2025-12-18-21-37-42.tar.gz

This will create a timestamped backup archive in the toolbox pod and copy it to the system where make target was run in the /tmp/eda-support/logs-<date> directory. The backup archive contains all the necessary information to restore your cluster.

Updating playground repository#

The workflow to upgrade EDA slightly differs depending on whether you have the original playground repository present in a system that you used to install EDA originally from or not.

git pull --rebase --autostash -v

git clone https://github.com/nokia-eda/playground && \
cd playground && \
make download-tools

edactl cluster

Name           Address  ActivityState  BuildVersion                  CoreVersion  AvgLatency(ms)  Reachable  Synchronized
 engine-config  self     Active         v25.4.1-2504252348-g720b7d2e  v2.0.0-0                     true       true

EDA_CORE_VERSION=25.4.1
EDA_APPS_VERSION=25.4.1

make download-pkgs

❯ make download-pkgs
--> INFO: Updating /home/rd/nokia-eda/playground/eda-kpt
--> INFO: Updating /home/rd/nokia-eda/playground/catalog
--> INFO: /home/rd/nokia-eda/playground/eda-kpt - selected version: 25.4.1
--> INFO: /home/rd/nokia-eda/playground/eda-kpt - is at 25.4.1
--> INFO: /home/rd/nokia-eda/playground/catalog - selected version: 25.4.1
--> INFO: /home/rd/nokia-eda/playground/catalog - is at 25.4.1

Ensure the package inventory is in sync with your existing cluster:

make cluster-restore-inventory

Uninstalling EDA components#

The existing EDA core components must be uninstalled, before installing the new version.

Breaking geo redundancy (optional)#

If you have a geo-redundant installation, on your active cluster member, update your EngineConfig to remove the .spec.cluster.redundant section. This will break the geo redundancy and allow you to upgrade the active member without affecting the standby member.

Stopping EDA platform#

To stop EDA components, enter the following command:

make eda-stop-core

This command returns no output, but will result in all Pods packaged as part of eda-kpt-base being stopped and removed from the cluster.

Uninstalling EDA core components#

Proceed with uninstalling EDA core components:

make eda-uninstall-core

You should now see no core components in your cluster. Check with the following command¹:

kubectl get pods -n eda-system

NAME                                  READY   STATUS    RESTARTS   AGE
cert-manager-csi-driver-n9bwk         3/3     Running   0          95m
eda-fluentbit-b7fns                   1/1     Running   0          96m
eda-fluentd-7cd48db9c5-9pvvp          1/1     Running   0          96m
eda-git-5db9dfc7bc-mn4rw              1/1     Running   0          95m
eda-git-replica-f69b9c9f4-bngf8       1/1     Running   0          95m
trust-manager-69955c46b8-bghj6        1/1     Running   0          95m

Stopping EDA git servers#

Continue with stopping EDA Git servers by scaling down the EDA Git deployments:

make scale-down-git-servers

Version-specific steps#

When upgrading from versions older than 25.12.1 stop the fluentd daemonset:

make uninstall-external-package-fluentd

Updating EDA kpt packages#

Set the desired EDA version in the prefs.mk file to match the target version you want to upgrade to. For example, to choose the 25.12.1 version:

EDA_CORE_VERSION=25.12.1
EDA_APPS_VERSION=25.12.1

Download the tools and packages by executing the following command:

make download-tools download-pkgs

Customizing kpt packages#

If you started with a freshly cloned repository, you want to add customizations to your EDA installation by setting the variables in the prefs.mk file; this process is explained in details at the installation phase.

At a minimum, ensure EXT_DOMAIN_NAME and EXT_HTTPS_PORT are set correctly in your prefs.mk file.

Configure the packages downloaded for the target EDA version with the customizations you added to the prefs.mk file:

make eda-configure-core

Installing the new version of EDA#

Install the new version of EDA core components by running:

make install-external-packages eda-install-core eda-is-core-ready

Restoring your backup#

Copy the backup file you collected at the beginning of this procedure from your machine back into the new eda-toolbox pod:

backupfile=/tmp/eda-support/logs-2025-12-18/eda-platform-backup-2025-12-18-21-37-42.tar.gz

toolboxpod=$(kubectl -n eda-system get pods \
-l eda.nokia.com/app=eda-toolbox -o jsonpath="{.items[0].metadata.name}")

kubectl -n eda-system cp $backupfile \
  $toolboxpod:/tmp/eda-backup.tar.gz

Restore your cluster to its previous state by running:

edactl platform restore /tmp/eda-backup.tar.gz

Upgrading your applications#

Installing the new version of EDA will deploy application versions according to the installed release; however, the backup restore operation will have restored application versions as they were set in the original cluster. These versions may be incompatible with the new version of EDA core, and must be upgraded immediately following the EDA backup restore. The existing Makefile can be used to do so:

make eda-install-apps

Resume NPP interactions#

To resume NPP interactions, set all TopoNode resources back to the normal mode.

make set-npp-mode-normal

After the script has been run, verify that the TopoNode resources are in normal mode:

kubectl get toponode -A \
-o custom-columns='NAMESPACE:.metadata.namespace,NAME:.metadata.name,MODE:.spec.npp.mode'

NAMESPACE       NAME     MODE
my-other-ns     leaf1    normal
my-other-ns     leaf2    normal
my-other-ns     leaf3    normal
my-other-ns     leaf4    normal
my-other-ns     spine1   normal
my-other-ns     spine2   normal
eda             leaf1    normal
eda             leaf2    normal
eda             spine1   normal

Verifying cluster health#

Check the following to ensure your cluster is healthy:

• All pods are running and healthy.
• All TopoNode resources are in normal mode, and have synced with their targets.
• No transaction failures exist.
• All cluster members are synchronized.