2023.10.5

Release date: August 14, 2024

New RHEL versions support

We have expanded our OS support to include RHEL 8.10 and 9.2, alongside existing support for 8.6, 8.8, and 8.9 versions.

AWS Signature Version 4 support

Automation Suite now uses AWS Signature Version 4 for all S3 API requests. This impacts the use of Automation Suite with S3 servers that do not support AWS Signature Version 4.

Optimized AI Center disk management

We have improved the configureUiPathDisks.sh script capabilities to give you more flexibility when managing AI Center storage. With this improvement, you can use the --aicenter-disk-name parameter to increase AI Center storage, whether you add a new disk or expand an existing one.

For more details, see Configuring the AI Center disk on selected machines.

Images cleanup for in-cluster Docker registry

You can now effortlessly remove obsolete images post-upgrade. This capability is specifically designed for offline setups with an in-cluster Docker registry. For more details, see Performing post-upgrade operations.

Security enhancements

We continue to provide security updates and patches to address Common Vulnerabilities and Exposures (CVEs).

Bug fixes

• Erratum - added February 27, 2025: We have fixed several bugs in AI Center when connected to an external Orchestrator and Identity Service. These bugs were specifically causing issues during setup in the following situations:
  • During the installation of Orchestrator certificates for trusted communication.
  • When setting up an Identity Service client for Document Manager.
  • When configuring an Identity Service client for metering services.
• We have fixed an issue causing upgrades from Automation Suite 2023.10.x to fail due to the Longhorn storage classes still being present after uninstalling Longhorn.
• For AI Center and Task Mining, the ListBucket API prerequisite check failed when using an external objectstore. However, AI Center and Task Mining do not use the ListBucket API permission. We have removed the unnecessary check.
• We have fixed an issue that prevented you from enabling SSO for ArgoCD due to a Dex image version discrepancy.
• We have addressed some issues in Apps, including certain operations failing in proxy environments and the legacy image control continuing to show a spinner after loading has completed.
• We have addressed an issue preventing some document validation tasks from being displayed accurately in Action Center.
• We have fixed an issue causing installation failures in certain conditions due to telemetry check errors. Previously, in case of failure, you had to manually set the telemetry check as optional and restart the installation. Now, a telemetry check failure only displays a warning and does not disrupt the installation process. This makes telemetry checks optional by default and ensures smooth, uninterrupted installations.
• We have fixed an issue causing the snapshot-controller-crds pod to remain in the CrashLoopBackOff state after an RKE2 upgrade. This issue occurred due to a conflict between the newly installed snapshot-controller and the existing one during the RKE2 upgrade.
• We have fixed an issue that caused the support bundle to not include historical logs and to not upload to the configured external object store. This issue occured in offline environments using an external OCI registry.
• In instances with host level SAML configured, when you accessed Automation Suite through a specific <Automation_Suite_URL>/<Organization_name> URL and selected the host level SAML login option, the ACS URL was incorrectly generated using the organization ID from the URL, instead of the expected host level ACS URL. We have fixed this. The system now correctly generates the host-level ACS URL during a host level SAML login, irrespective of the specific organization URL used to access Automation Suite.
• Previously, you could not save your SMTP settings without providing a username and password, even though these details were optional. This behavior no longer occurs.

Known issues

FIPS 140-2 support limitations

Erratum - added January 22, 2026: Insights is not supported on Automation Suite deployments that run on FIPS 140-2-enabled machines. To remain compliant with FIPS 140-2 requirements, you must disable Insights.

For details, refer to Security and compliance.

Kerberos keytab rotation does not trigger token regeneration

Erratum - added September 24, 2025: An issue causes Kerberos keytab rotation to not immediately regenerate authentication tokens, which may lead to temporary connectivity disruptions between services and the database.

To address the issue, you must manually trigger the Kerberos ticket renewal cronjob by running the following command:

kubectl delete job tgt-rotate-manual -n uipath --ignore-not-found ; kubectl create job tgt-rotate-manual --from=cronjob/kerberos-tgt-update -n uipath

We fixed the issue in Automation Suite 2023.10.11.

GPU enablement fails with external registries containing project names

Erratum - added September 24, 2025: An issue prevents GPU enablement after adding a GPU node when using external registries with project names (such as Harbor). The required pods do not start and display an ImagePullBackoff error.

If you are using an external registry with a project name, update the NVIDIA device plugin daemonset with the following command:

# Replace REGISTRY_WITH_PROJECT_NAME with the correct value (Eg. my.registry.io:443/myproject)
# Replcae TAG with the correct value. You can get this from the <installer_directory>/versions/docker-images.json file (Eg. v0.17.1)
kubectl set image daemonset/nvidia-device-plugin-daemonset \
  -n kube-system \
  nvidia-device-plugin-ctr=<REGISTRY_WITH_PROJECT_NAME>/k8s-device-plugin:<TAG>

We fixed the issue in Automation Suite 2023.10.11.

Orchestrator does not start when using user-assigned managed identity

Erratum - added September 24, 2025: An issue prevents the storage.useClientID parameter from being set when you select a user-assigned managed identity for object storage. As a result, Orchestrator cannot start in environments where access is restricted to user-assigned managed identities.

To address the issue, you must manually set the parameter in ArgoCD, as follows:

1. In the ArgoCD Orchestrator application, go to Details > Parameters.
2. In the values text box, set storage.isExternal : true and save.

We fixed the issue in Automation Suite 2023.10.11.

Thanos compactor stops on corrupted blocks

Erratum - added September 24, 2025: An issue causes the Thanos compactor to stop compacting metrics when it encounters corrupted blocks in the object store. This prevents compaction and leads to increased storage usage.

To address the issue, refer to the Troubleshooting section.

We fixed the issue in Automation Suite 2023.10.11.

Task Mining Kerberos initialization failure

Erratum - added June 26, 2025: An issue causes the Task Mining service to fail to initialize in Kerberos environments. This issue occurs when the kerberos-discoverygroup-user-keytab secret is missing or when the discoverygroup app is not included in the enabledApps parameter of the IntegratedAuth app.

To address the issue, refer to the Troubleshooting section.

We fixed the issue in Automation Suite 2023.10.10.

Agent node addition failure in offline environments

Erratum - added June 26, 2025: An issue causes the addition process of agent nodes in offline environments to fail. Agent node additions require the temporary registry to be running. To address the issue, you must re-enable the temporary registry to successfully join the agent node.

We fixed the issue in Automation Suite 2023.10.10.

Installation fails due to CA certificates with large policy OIDs

Erratum - added June 26, 2025: An issue causes the Automation Suite installation to fail when using Certificate Authority (CA) certificates that include a CertificatePolicies section with policy OID values exceeding 4 bytes. This issue prevents Automation Suite from recognizing these CA certificates, interrupting the installation process.

To address the issue, you must either ensure that the policy OID values within the CertificatePolicies section do not exceed 4 bytes, or avoid using CA generators that add the CertificatePolicies section to the certificates.

We fixed the issue in Automation Suite 2023.10.10.

Licensing SQL connection error

Erratum - added June 26, 2025: Licensing SQL connection errors occur when the Data Source property specified both a named instance and a port.

We fixed the issue in Automation Suite 2023.10.10.

Grafana dashboard error

Erratum - added June 10, 2025: In multi-node (HA) setups, Grafana runs with two replicas, but there is no common data source between the replicas, causing issues in user session management. Errors like Dashboard not found and Invalid dashboard UID in the request appear in the Grafana dashboard.

As a workaround, since Grafana is not a critical application, you can reduce the replica count to one by performing the following steps:

1. Modify the helm parameter for the fabric-installer application. This automatically triggers the sync operation for fabric-installer application.

   kubectl patch   application fabric-installer -n argocd    --type='json'  -p='[{"op": "add", "path": "/spec/source/helm/parameters/-", "value": {"name": "global.rancherMonitoring.grafana.replicas", "value": "1"}}]'
   

2. Sync the rancher-monitoring application from ArgoCD UI.

We have fixed this issue in Automation Suite 2023.10.7.

uipathctl health check command

Erratum - added May 21, 2025: For offline Automation Suite installations using the external docker registry, the uipathctl health check command does not work due to a known issue. This issue has been fixed in Automation Suite 2023.10.6.

As a workaround, you can download the uipathctl binary from Automation Suite 2023.10.6 and use it to run the uipathctl health check command.

Kubernetes upgrade failure

Erratum - added April 15, 2025: An issue causes the Kubernetes upgrade to fail due to a missing certificate-injector-webhook in deployment.apps. To address the issue, you must run the following command:

kubectl delete configmap -n uipath-infra certificate-injector-webhook-configmap --ignore-not-found

We fixed the issue in Automation Suite 2023.10.9.

Document Understanding deployment issue post-upgrade

Erratum - added April 15, 2025: An issue prevents you from importing files into your DM session in AI Center after upgrading to Automation Suite 2023.10.x. This issue occurs due to the du-digitizer-worker deployment from the old Automation Suite version not being fully cleaned up after the upgrade. Starting with Automation Suite 2023.10.0, the deployment was renamed to du-digitizer-worker-deployment.

To address the issue, you must navigate to the ArgoCD UI and from the documentunderstanding application, manually delete the old du-digitizer-worker deployment. After deletion, re-sync the documentunderstanding application.

We fixed the issue in Automation Suite 2023.10.8.

Incomplete Docker registry cleanup process

Erratum - added April 15, 2025: An issue prevents the registry cleanup command from effectively clearing the unused Docker images from all registry pods. To address the issue, refer to the Troubleshooting section.

We fixed the issue in Automation Suite 2023.10.9.

Upgrade fails due to MongoDB to SQL Server migration

Erratum - added April 15, 2025: We have identified an issue impacting the side-by-side and the in-place upgrade of Automation Suite 2022.10 or earlier to Automation Suite 2023.10.0 or later. Due to a faulty migration from MongoDB to SQL Server, the upgrade operation fails if you have Apps enabled and use Kerberos authentication for the SQL Server database.

The recommended solution is to upgrade to Automation Suite 2023.10.9.

Kube-proxy health check not working

Erratum - added April 15, 2025: The node-monitor component, which monitors the node for issues such as checking kube-proxy health or if ip_forward is enabled, and cordons the node if issues arise, is not working for specific nodes such as GPU, task-mining, or as-robot. We have fixed this issue in Automation Suite 2023.10.9.

Kerberos tickets do not automatically renew

Erratum - added April 15, 2025: Kerberos tickets fail to automatically renew after cluster restoration, requiring manual sync of the Windows Auth app.

This issue is fixed in Automation Suite 2023.10.9.

AI Center storage migration failure

Erratum - added February 13, 2025: An issue prevents uipathctl from correctly setting executeMigration to true during the migration process from Ceph in-cluster storage to external object storage in AI Center.

To fix this issue, you must manually set executeMigration to true in ArgoCD for AI Center and then proceed with a resync, ensuring to prune. This issue is fixed in Automation Suite 2023.10.8.

ML skill deployment failure due to token expiry

Erratum - added February 13, 2025: Due to a containerd issue, the deployment of ML skills fails as a token expires before larger image downloads can be completed, especially under conditions of slower network connectivity.

To address this isssue, see the Troubleshooting section.

Email alerts configuration failure post upgrade

Erratum - added February 13, 2025: An issue affects the functionality of email alerts configuration and the setup of multiple receivers after upgrading from Automation Suite 2022.10. To address this issue, see Troubleshooting.

This issue is fixed in Automation Suite 2023.10.8.

Data loss when reinstalling or upgrading Insights following Automation Suite upgrade

Erratum - added December 18, 2024: Following an Automation Suite upgrade from version 2023.4 or older, reinstalling or upgrading Insights results in data loss due to an issue with Insights storage class changes. To address the problem, see Troubleshooting.

We fixed the issue in Automation Suite 2023.10.7.

Insights annotation issue blocks installer

Erratum - added December 18, 2024: An Insights annotation issue blocks the Automation Suite installer. We fixed the issue in Automation Suite 2023.10.7.

Node maintenance issue in non-HA Redis

Erratum - added December 18, 2024: An issue causes the redis-cluster-0 pod to get stuck in the terminating state when you perform a node drain in non-HA Redis scenarios. To address the problem, you must force delete the pod using the following command:

kubectl -n redis-system delete pod redis-cluster-0 --force

We fixed the issue in Automation Suite 2023.10.7.

RKE2 log collector runs indefinitely

Erratum - added December 18, 2024: The RKE2 log collector runs indefinitely for some system commands. We fixed the issue in Automation Suite 2023.10.7.

AI Center skills sync failure during side-by-side upgrade

Erratum - added December 18, 2024: A syntax issue causes the failure of the aicenter skill sync and aicenter skill status commands during a side-by-side upgrade.

To fix this issue, you must manually edit the syntax. For details, see the Troubleshooting section.

Unnecessary Redis DNS resolution validations during Active/Passive checks

Erratum - added December 18, 2024: An issue causes unnecessary Redis DNS resolution validations for redis.<primary-fqdn> and redis-db.<primary-fqdn> during Active/Passive prerequisite checks. This issue is fixed in Automation Suite 2023.10.7.

Upgrade failure due to overridden Insights PVC sizes

Erratum - added December 18, 2024: An issue causes upgrades to fail when the existing Insights PVC sizes are inadvertently overridden. To address this problem, you must manually change the PVC sizes in ArgoCD UI. For details, see the Troubleshooting section.

This issue is fixed in Automation Suite 2023.10.7.

Unintended RKE2 service upgrade on additional nodes

Erratum - added November 26, 2024: We have identified an issue where exclude= rke2-* is not added to the /etc/yum.conf file on nodes other than the first server. In specific environments, particularly online ones, an attempt to upgrade all components can cause an unintentional upgrade of the RKE2 service on nodes other than the first server.

To fix this issue, you must manually add exclude=rke2-* to the /etc/yum.conf file on all the nodes of your Automation Suite cluster.

Test Automation SQL connection string is ignored

Erratum - added October 17, 2024: When you provide an SQL connection string under the orchestrator.testautomation section of the cluster_config.json file, the uipathctl binary ignores the connection string and uses the one under the orchestrator section instead. To address the problem, see the Troubleshooting section.

OSD PV resize does not persist after upgrade

If you resize the rook-ceph OSD PV, the new size does not persist following an Automation Suite upgrade.

Installation fails while populating node labels

Erratum - added October 17, 2024: When deploying Automation Suite on AWS machines where only IMDSv2 is enabled, installation fails while populating node labels. To address the issue, see the Important note in Optional: Enabling resilience to zonal failures in a multi-node HA-ready production cluster.

Cannot upgrade due to failed jobs in system-upgrade namespace

Erratum - Added October 17, 2024: We accidentally documented this issue as fixed in Automation Suite 2023.10.5. However, the behavior persists. We fixed the issue in Automation Suite 2023.10.6.

Upgrading Automation Suite fails due to the presence of failed jobs in the system-upgrade namespace. If the upgrade command fails at any stage (infra, fabric, or service upgrade), take the following steps before retrying the upgrade:

1. List the existing jobs in the system-upgrade namespace:

   kubectl get jobs -n system-upgrade
   

2. Delete the failed jobs:

   kubectl -n system-upgrade delete jobs <failed_jobs>
   

When running the command, replace the <failed_jobs> placeholder with the names of the failed jobs, separated by spaces.

Pods cannot communicate with FQDN in a proxy environment

Erratum - added October 17, 2024: 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. The issue causes the following error:

System.Net.Http.HttpRequestException: The proxy tunnel request to proxy 'http://<proxyFQDN>:8080/' failed with status code '404'.

To fix the issue, see the Troubleshooting section.

AI Center issues with external Orchestrator and Identity Service

Erratum - added September 11, 2024: There are several bugs in AI Center when connected to an external Orchestrator and Identity Service, causing issues during setup in the following situations:

• During the installation of Orchestrator certificates for trusted communication.
• When setting up an Identity Service client for Document Manager.
• When configuring an Identity Service client for metering services.

Weak ciphers in TLS 1.2

Erratum - added August 29, 2024: We have identified certain vulnerabilities associated with the usage of weak ciphers in TLS 1.2. For details on how to mitigate the issue, see How to address weak ciphers in TLS 1.2.

Intermittent not ready reports from agent machines

Erratum - added October 17, 2024: An issue causes agent machines to intermittently report as not ready when the rke2-server service is shut down without executing the rke2-killall.sh script. We fixed this issue in Automation Suite 2023.10.6.

Registry configuration corruption during upgrades

Erratum - added October 17, 2024: An issue causes the corruption of the registry configuration during the upgrade process. The issue occurs when uipathctl mistakenly processes ports in helm URLs as external registries, excluding the insecure_skip_verify: true parameter from the registries.yaml file. To overcome this issue, you must manually include the parameter in your registries.yaml file. We fixed this issue in Automation Suite 2023.10.6.

Resized PVC upgrade issue

Erratum - added October 17, 2024: An issue is causing problems in the upgrade process if you have resized the Docker-Registry PVC used by your in-cluster Docker registry. Specifically, the resized PVC is not accurately recognized during upgrades. We fixed this issue in Automation Suite 2023.10.6.

Redis failures due to inconsistent password updates

Erratum - added October 17, 2024: An issue is causing Redis recovery job failures when you update the ArgoCD admin password directly through the UI without simultaneously updating the Kubernetes secrets as well.

To prevent the issue, we highly recommend that each time you change the ArgoCD admin password via the UI, ensure that you update the Kubernetes secrets at the same time. For more details on how to properly update the ArgoCD admin password, refer to Troubleshooting.

Enabling Connaisseur causes installation or upgrade failures

Erratum - added October 17, 2024: An issue causes installation or upgrade failures when enabling Connaisseur during the configuration of an external OCI-compliant registry. The isssue occurs when you set the registries.trust.enabled parameter to true in the cluster_config.json file.

Deprecation timeline

We recommend that you regularly check the deprecation timeline for any updates regarding features that will be deprecated and removed.

Bundling details

Product versions

To find out what has changed on each Automation Suite product, visit the following links.

If the product is greyed out, this new Automation Suite version does not bring any changes to it.

DISCOVER	BUILD	MANAGE	ENGAGE
Automation Hub 2023.10.5	Automation Ops 2023.10.5	AI Center 2023.10.5	Action Center 2023.10.5
Task Mining 2023.10.5	AI Computer Vision 2023.10.4	Insights 2023.10.5	Apps 2023.10.5
Process Mining 2023.10.5	Document Understanding 2023.10.5	Orchestrator 2023.10.6
		Test Manager 2023.10.5
		Data Service 2023.10.5

Internal third-party component versions

This Automation Suite release bundles the following internal components:

Component	Version
RKE2	1.28.7
ArgoCD	2.10.9
logging-operator	4.6.0
logging-operator-logging	4.6.0
gatekeeper	3.16.0
rook-ceph	1.9.4
prometheus-pushgateway	2.12.0
cert-manager	1.14.5
rancher-istio	103.3.0-up1.21.1
rancher-logging	103.0.0-up3.17.10
rancher-logging-crd	103.0.0-up3.17.10
rancher-monitoring-crd	103.1.0-up45.31.1
rancher-gatekeeper-crd	103.1.0-up3.13.0
rancher-gatekeeper	103.1.0-up3.13.0
rancher-monitoring	103.1.0-up45.31.1
longhorn	1.5.5
longhorn-crd	1.1.100
reloader	1.0.95
csi-driver-smb	1.14.0
velero	6.2.0
redis-operator	7.4.2-12
redis-cluster	7.4.2-129

Migration tool version

The migration tool version you need depends on the standalone products you plan to migrate and the targeted Automation Suite version. For more details, see Migration compatibility matrix.

For instructions on migrating a standalone product to the current version of Automation Suite, see Full migration.