- Overview
- Requirements
- Installation
- Post-installation
- Migration and upgrade
- Upgrading Automation Suite on EKS/AKS
- Step 1: Moving the Identity organization data from standalone to Automation Suite
- Step 2: Restoring the standalone product database
- Step 3: Backing up the platform database in Automation Suite
- Step 4: Merging organizations in Automation Suite
- Step 5: Updating the migrated product connection strings
- Step 6: Migrating standalone Orchestrator
- Step 7: Migrating standalone Insights
- Step 8: Migrating standalone Test Manager
- Step 9: Deleting the default tenant
- Performing a single tenant migration
- Migrating from Automation Suite on Linux to Automation Suite on EKS/AKS
- Monitoring and alerting
- Cluster administration
- Product-specific configuration
- Troubleshooting
- The backup setup does not work due to a failure to connect to Azure Government
- Pods in the uipath namespace stuck when enabling custom node taints
- Unable to launch Automation Hub and Apps with proxy setup
- Pods cannot communicate with FQDN in a proxy environment
- Test Automation SQL connection string is ignored
- EKS backup failure due to Velero version
- Velero backup fails with FailedValidation error
- Accessing FQDN returns RBAC access denied error
- Manual ArgoCD NetworkPolicy mitigation (GHSA-47m3-95c7-g2g8)
Proxy
Reference for proxy configuration options and allowlist requirements for deploying Automation Suite on EKS/AKS in a proxy environment.
To deploy Automation Suite in a proxy configuration, you must configure your cluster and add all UiPath® required domains to an allowlist in your network proxy.
A proxy configuration is required only if you perform an online installation.
Automation Suite does not support the IPv6 internet protocol.
Configuring the cluster
For instructions on how to configure the cluster worker nodes in proxy mode, refer to the Microsoft or AWS documentation:
- AWS - Automate HTTP proxy configuration for Amazon EKS containerd nodes
- Microsoft - Configuring Azure Kubernetes Service (AKS) nodes with an HTTP proxy - Azure Kubernetes Service
When configuring the cluster nodes in proxy mode, make sure to add the following domains to the no_proxy list:
- EKS
"$VPC_CIDR,$SERVICE_CIDR, localhost,127.0.0.1,169.254.169.254,<clusterfqdn>,.<clusterfqdn>,.internal,.eks.amazonaws.com,.cloudfront.net,s3.CLOUD_REGION.amazonaws.com,.s3.CLOUD_REGION.amazonaws.com,.dkr.ecr.CLOUD_REGION.amazonaws.com,ec2.CLOUD_REGION.amazonaws.com,api.ecr.CLOUD_REGION.amazonaws.com,.elb.amazonaws.com,.gr7.CLOUD_REGION.eks.amazonaws.com,.s3.amazonaws.com,kubernetes,kubernetes.default,kubernetes.default.svc,kubernetes.default.svc.cluster,kubernetes.default.svc.cluster.local,.svc,.svc.cluster,.svc.cluster.local,.svc.cluster.local.,argocd-repo-server,istiod.istio-system.svc,logging-operator-logging-fluentd.logging.svc.cluster.local,argocd-repo-server,.local,.cluster,ai-helper-svc,ai-pkgmanager-svc,ai-deployer-svc,ai-appmanager-svc,ai-trainer-svc,get""$VPC_CIDR,$SERVICE_CIDR, localhost,127.0.0.1,169.254.169.254,<clusterfqdn>,.<clusterfqdn>,.internal,.eks.amazonaws.com,.cloudfront.net,s3.CLOUD_REGION.amazonaws.com,.s3.CLOUD_REGION.amazonaws.com,.dkr.ecr.CLOUD_REGION.amazonaws.com,ec2.CLOUD_REGION.amazonaws.com,api.ecr.CLOUD_REGION.amazonaws.com,.elb.amazonaws.com,.gr7.CLOUD_REGION.eks.amazonaws.com,.s3.amazonaws.com,kubernetes,kubernetes.default,kubernetes.default.svc,kubernetes.default.svc.cluster,kubernetes.default.svc.cluster.local,.svc,.svc.cluster,.svc.cluster.local,.svc.cluster.local.,argocd-repo-server,istiod.istio-system.svc,logging-operator-logging-fluentd.logging.svc.cluster.local,argocd-repo-server,.local,.cluster,ai-helper-svc,ai-pkgmanager-svc,ai-deployer-svc,ai-appmanager-svc,ai-trainer-svc,get" - AKS
"<cluster-fqdn>,.<cluster-fqdn>,localhost,127.0.0.1,10.0.0.0/8,mcr.microsoft.com,kubernetes,kubernetes.default,kubernetes.default.svc,kubernetes.default.svc.cluster,kubernetes.default.svc.cluster.local,.svc,.svc.cluster,.svc.cluster.local,.svc.cluster.local.,argocd-repo-server,istiod.istio-system.svc,logging-operator-logging-fluentd.logging.svc.cluster.local,argocd-repo-server,.local,.cluster,ai-helper-svc,ai-pkgmanager-svc,ai-deployer-svc,ai-appmanager-svc,ai-trainer-svc,get\""<cluster-fqdn>,.<cluster-fqdn>,localhost,127.0.0.1,10.0.0.0/8,mcr.microsoft.com,kubernetes,kubernetes.default,kubernetes.default.svc,kubernetes.default.svc.cluster,kubernetes.default.svc.cluster.local,.svc,.svc.cluster,.svc.cluster.local,.svc.cluster.local.,argocd-repo-server,istiod.istio-system.svc,logging-operator-logging-fluentd.logging.svc.cluster.local,argocd-repo-server,.local,.cluster,ai-helper-svc,ai-pkgmanager-svc,ai-deployer-svc,ai-appmanager-svc,ai-trainer-svc,get\"
Make sure to replace <cluster-fqdn> with the actual FQDN URL.
Providing proxy setting in input.json
Configure the input.json file with the following parameters:
| Mandatory parameters | Description |
|---|---|
enabled | Use true or false to enable or disable proxy settings. |
http_proxy | Used to route HTTP outbound requests from the cluster. This should be the proxy server FQDN and port. |
https_proxy | Used to route HTTPS outbound requests from the cluster. This should be the proxy server FQDN and port. |
no_proxy | Comma-separated list of hosts, IP addresses, or IP ranges in CIDR format that you do not want to route via the proxy server. The list must be the one provided in the Configuring the cluster section for EKS or AKS, respectively. |
Example input.json for proxy configuration:
"proxy": {
"enabled": true,
"http_proxy": "http://<PROXY-SERVER-IP:<PROXY-PORT>",>
"https_proxy": "http://<PROXY-SERVER-IP:<PROXY-PORT>",>
"no_proxy": <"paste list from Configuring the cluster section">
}
"proxy": {
"enabled": true,
"http_proxy": "http://<PROXY-SERVER-IP:<PROXY-PORT>",>
"https_proxy": "http://<PROXY-SERVER-IP:<PROXY-PORT>",>
"no_proxy": <"paste list from Configuring the cluster section">
}
Adding URLs to an allowlist
URLs for EKS and AKS
.microsoft.com
login.microsoftonline.com
login.windows.net
graph.microsoft.com
sfbrprddeploywe.azurecr.io
sfbrprddeploywe.westeurope.data.azurecr.io
registry-data.uipath.com
registry.uipath.com
dc.services.visualstudio.com
activate.uipath.com
download.uipath.com
du-metering.uipath.com
du.uipath.com
du-au.uipath.com
du-ca.uipath.com
du-jp.uipath.com
du-us.uipath.com
du-sg.uipath.com
raw.githubusercontent.com
objects.githubusercontent.com
pkg-containers.githubusercontent.com
raw.github.com
api.github.com
api.nuget.org
pkgs.dev.azure.com
gallery.uipath.com
github.com
pypi.org
pypi.python.org
pythonhosted.org
files.pythonhosted.org
opensuse.org
copr.fedorainfracloud.org
download.copr.fedorainfracloud.org
ping.looker.com
rt.services.visualstudio.com
uipath.pkgs.visualstudio.com
.blob.core.windows.net
pkgs.dev.azure.com
events.launchdarkly.com
app.launchdarkly.com
in.applicationinsights.azure.com
.pkg.dev
production.cloudflare.docker.com
.googleapis.com
.microsoft.com
login.microsoftonline.com
login.windows.net
graph.microsoft.com
sfbrprddeploywe.azurecr.io
sfbrprddeploywe.westeurope.data.azurecr.io
registry-data.uipath.com
registry.uipath.com
dc.services.visualstudio.com
activate.uipath.com
download.uipath.com
du-metering.uipath.com
du.uipath.com
du-au.uipath.com
du-ca.uipath.com
du-jp.uipath.com
du-us.uipath.com
du-sg.uipath.com
raw.githubusercontent.com
objects.githubusercontent.com
pkg-containers.githubusercontent.com
raw.github.com
api.github.com
api.nuget.org
pkgs.dev.azure.com
gallery.uipath.com
github.com
pypi.org
pypi.python.org
pythonhosted.org
files.pythonhosted.org
opensuse.org
copr.fedorainfracloud.org
download.copr.fedorainfracloud.org
ping.looker.com
rt.services.visualstudio.com
uipath.pkgs.visualstudio.com
.blob.core.windows.net
pkgs.dev.azure.com
events.launchdarkly.com
app.launchdarkly.com
in.applicationinsights.azure.com
.pkg.dev
production.cloudflare.docker.com
.googleapis.com
Additional URLs for EKS
.s3.dualstack.CLOUD_REGION.amazonaws.com
public.ecr.aws
.cloudfront.net
.s3.dualstack.CLOUD_REGION.amazonaws.com
public.ecr.aws
.cloudfront.net
Pods failing to communicate with FQDN in a proxy environment
In a proxy environment, if the proxy server uses the same port as the TCP port of any other service in the Istio service mesh, such as port 8080, pods cannot communicate with the FQDN. For more information on the issue and how to address it, refer to Pods cannot communicate with FQDN in a proxy environment.