ocp_config¤
Configure OpenShift Container Platform cluster-level settings for optimal MAS deployment. This role provides essential tuning for ingress timeouts, TLS cipher compatibility with IBM Java Semeru FIPS mode, and OperatorHub catalog source management.
Key capabilities:
- Ingress timeout tuning: Prevent request failures for long-running operations
- FIPS-compatible TLS ciphers: Enable IBM Java Semeru runtime in FIPS mode
- Catalog source management: Disable default Red Hat catalogs for air-gapped environments
- Path-based routing: Configure namespace ownership for multi-namespace route support
This role configures:
- Tune the
IngressControllerto avoid request failures due to timeout for long running requests - Configure the
IngressControllernamespace ownership for path-based routing support - Update
APIServerandIngressControllerto set a customtlsSecurityProfileto accommodate ciphers supported by IBM Java Semeru runtime. This is required for allowing the Java applications using Semeru runtime to run in FIPS mode. The following ciphers will be enabled: TLS_AES_128_GCM_SHA256TLS_AES_256_GCM_SHA384TLS_CHACHA20_POLY1305_SHA256ECDHE-ECDSA-AES128-GCM-SHA256ECDHE-RSA-AES128-GCM-SHA256ECDHE-ECDSA-AES256-GCM-SHA384ECDHE-RSA-AES256-GCM-SHA384ECDHE-ECDSA-CHACHA20-POLY1305ECDHE-RSA-CHACHA20-POLY1305DHE-RSA-AES128-GCM-SHA256DHE-RSA-AES256-GCM-SHA384ECDHE-RSA-AES128-SHA256ECDHE-RSA-AES128-SHAECDHE-RSA-AES256-SHA- Disable the default Red Hat
CatalogSources: certified-operatorscommunity-operatorsredhat-operators
Role Variables¤
| Variable | Required | Default |
|---|---|---|
ocp_update_ciphers_for_semeru |
No | false |
ocp_ingress_controller_name |
No | default |
ocp_ingress_update_timeouts |
No | false |
ocp_ingress_client_timeout |
No | 30s |
ocp_ingress_server_timeout |
No | 30s |
ocp_ingress_namespace_ownership |
No | Not set |
ocp_operatorhub_disable_redhat_sources |
No | false |
ocp_enable_ipv6 |
No | false |
cluster_type |
No | None |
cluster_name |
No | Not set |
rosa_token |
Conditional | None |
ocp_update_allowed_registries |
No | false |
ocp_allowed_registries |
No | [] |
ocp_add_mas_registries |
No | true |
ocp_add_rh_registries |
No | true |
ocp_update_additional_trusted_ca |
No | false |
ocp_additional_trusted_ca_dir |
No | /tmp |
rosa_token is only required when cluster_type=rosa-hcp; the allowed registries and additional trusted CA settings currently work with ROSA HCP clusters only.
API Server¤
ocp_update_ciphers_for_semeru¤
Apply a custom TLS security profile — including ciphers compatible with the IBM Java Semeru runtime — to both the APIServer and IngressController resources. Required when running MAS applications (particularly Manage) in FIPS mode with IBM Java Semeru runtime.
- Optional
- Environment Variable:
OCP_UPDATE_CIPHERS_FOR_SEMERU - Default:
false
When enabled, the tlsSecurityProfile on both resources is updated to enable TLS 1.3 ciphers TLS_AES_128_GCM_SHA256, TLS_AES_256_GCM_SHA384, and TLS_CHACHA20_POLY1305_SHA256, plus TLS 1.2 ciphers ECDHE-ECDSA-AES128-GCM-SHA256, ECDHE-RSA-AES128-GCM-SHA256, ECDHE-ECDSA-AES256-GCM-SHA384, ECDHE-RSA-AES256-GCM-SHA384, ECDHE-ECDSA-CHACHA20-POLY1305, ECDHE-RSA-CHACHA20-POLY1305, DHE-RSA-AES128-GCM-SHA256, DHE-RSA-AES256-GCM-SHA384, ECDHE-RSA-AES128-SHA256, ECDHE-RSA-AES128-SHA, and ECDHE-RSA-AES256-SHA.
Warning
Changes affect the cluster-wide TLS configuration, may require a cluster restart to fully apply, and should be verified for compatibility with other applications in the cluster.
Ingress Controller¤
ocp_ingress_controller_name¤
Name of the Ingress Controller to configure — default (the default cluster Ingress Controller, most common) or any custom Ingress Controller name in your cluster.
- Optional
- Environment Variable:
OCP_INGRESS_CONTROLLER_NAME - Default:
default
Both the timeout settings (ocp_ingress_update_timeouts) and the namespace ownership policy (ocp_ingress_namespace_ownership) are applied to this controller. Custom Ingress Controllers are rare and typically used for advanced routing scenarios — verify the controller name exists before configuring.
ocp_ingress_update_timeouts¤
Whether to apply custom client and server timeout values to the Ingress Controller — enable when experiencing timeout issues with long-running MAS operations such as large file uploads, report generation, or data imports.
- Optional
- Environment Variable:
OCP_INGRESS_UPDATE_TIMEOUTS - Default:
false
When enabled, the IngressController is updated with the values from ocp_ingress_client_timeout and ocp_ingress_server_timeout. The default OpenShift timeout (30s) may be insufficient for MAS Manage operations — enabling this is recommended for production MAS deployments. Changes apply to all routes in the cluster.
ocp_ingress_client_timeout¤
Client-side timeout duration for ingress connections — how long a connection is held open while waiting for a client to send data or complete a request. A duration string with unit suffix (e.g. 30s, 5m, 1h).
- Optional
- Environment Variable:
OCP_INGRESS_CLIENT_TIMEOUT - Default:
30s
Common values: 30s (default, suitable for most operations), 5m (recommended for MAS Manage file uploads), 10m (very large file operations). Only applies when ocp_ingress_update_timeouts is true; affects all routes in the cluster. Consider network latency and file sizes when setting — too short may cause upload failures, too long may consume resources.
ocp_ingress_server_timeout¤
Server-side timeout duration for ingress connections — how long a connection is held open while waiting for a backend application to respond. A duration string with unit suffix (e.g. 30s, 5m, 1h).
- Optional
- Environment Variable:
OCP_INGRESS_SERVER_TIMEOUT - Default:
30s
Common values: 30s (default, suitable for most operations), 5m (recommended for MAS Manage report generation), 10m (complex data processing operations), 30m (very long-running batch operations). Only applies when ocp_ingress_update_timeouts is true; affects all routes in the cluster — monitor application logs to determine appropriate values.
ocp_ingress_namespace_ownership¤
Namespace ownership policy for the Ingress Controller — set to InterNamespaceAllowed to enable path-based routing support, which allows routes to claim the same hostname across different namespaces.
- Optional
- Environment Variable:
OCP_INGRESS_NAMESPACE_OWNERSHIP - Default: Not set
Note
When both timeout settings and namespace ownership are configured, they are applied in a single atomic operation to the IngressController resource.
OperatorHub¤
ocp_operatorhub_disable_redhat_sources¤
Whether to disable the default Red Hat OperatorHub catalog sources (certified-operators, community-operators, redhat-operators) — required for air-gapped/disconnected environments where external catalog sources are not accessible, and useful to enforce the use of mirrored catalogs only.
- Optional
- Environment Variable:
OCP_OPERATORHUB_DISABLE_REDHAT_SOURCES - Default:
false
When true, the three default Red Hat catalog sources are disabled and operators must be installed from custom or mirrored catalogs — ensure custom catalogs are configured before disabling the defaults. Affects all operator installations in the cluster.
Warning
Setting to false does not re-enable disabled sources. Operators cannot be installed from Red Hat catalogs after disabling.
IPv6 Networking¤
ocp_enable_ipv6¤
Enable dual-stack IPv4/IPv6 networking on the cluster — adds an IPv6 cluster network (fd01::/48) and service network (fd02::/112), then waits for all cluster operators and nodes to become healthy, verifies all workers have IPv6 configured, and approves any pending node CSRs.
- Optional
- Environment Variable:
OCP_ENABLE_IPV6 - Default:
false
Only required for Fyre clusters at the RTP (Research Triangle Park) site — SVL (San Jose/Silicon Valley, the default site) and other sites use IPv4.
ROSA-HCP Settings¤
cluster_type¤
Deployment model of OpenShift — currently only used for rosa-hcp (AWS ROSA HCP), which triggers the deployment-model-specific tasks such as updating allowed registries and additional trusted CAs.
- Optional
- Environment Variable:
CLUSTER_TYPE - Default: None
cluster_name¤
Name of the OpenShift cluster — used as the target cluster for rosa CLI commands.
- Optional
- Environment Variable:
CLUSTER_NAME - Default: Not set
Currently only used for rosa-hcp.
rosa_token¤
ROSA login token for the rosa CLI, authenticating it against the ROSA cluster.
- Required (when
cluster_type=rosa-hcp) - Environment Variable:
ROSA_TOKEN - Default: None
Allowed Registries¤
ocp_update_allowed_registries¤
Whether to update the cluster's allowed registries list, restricting image pulls to only the defined registries — required for air-gapped/disconnected environments.
- Optional
- Environment Variable:
OCP_UPDATE_ALLOWED_REGISTRIES - Default:
false
When true, the allowed registries are set from ocp_allowed_registries, optionally extended by ocp_add_mas_registries and ocp_add_rh_registries; all other registries are blocked. ROSA HCP automatically adds the Red Hat required registries image-registry.openshift-image-registry.svc:5000, quay.io, and registry.redhat.io.
Warning
Currently works with ROSA HCP only. Setting to false does not remove the allowed registries list.
ocp_allowed_registries¤
List of allowed registries to add to the cluster, beyond those defined by MAS and Red Hat — for air-gapped/disconnected environments, provide the domain(s) of your own image registry or registries.
- Optional
- Environment Variable:
OCP_ALLOWED_REGISTRIES - Default:
[]
Pods in the cluster can pull images only from the registries in the defined list.
Warning
Currently works with ROSA HCP only. Setting to [] does not remove the allowed registries list. When exporting OCP_ALLOWED_REGISTRIES, use a comma-separated string — for example export OCP_ALLOWED_REGISTRIES=mas.ibm.com,abc.com.
ocp_add_mas_registries¤
Whether to add the MAS-specific registries icr.io, cp.icr.io, quay.io, and gcr.io on top of those defined in ocp_allowed_registries.
- Optional
- Environment Variable:
OCP_ADD_MAS_REGISTRIES - Default:
true
Warning
Currently works with ROSA HCP only.
ocp_add_rh_registries¤
Whether to add the Red Hat-specific registries icr.io, registry.redhat.io, registry.connect.redhat.com, ghcr.io, docker.io, nvcr.io, and gcr.io on top of those defined in ocp_allowed_registries.
- Optional
- Environment Variable:
OCP_ADD_RH_REGISTRIES - Default:
true
Warning
Currently works with ROSA HCP only.
Additional Trusted CA¤
ocp_update_additional_trusted_ca¤
Whether to update the cluster's additional trusted CAs — adding the CAs of the image registries to be used so that the cluster can trust them. Required for air-gapped/disconnected environments.
- Optional
- Environment Variable:
OCP_UPDATE_ADDITIONAL_TRUSTED_CA - Default:
false
Warning
Currently works with ROSA HCP only, and setting to false does not remove additional trusted CAs. If ocp_additional_trusted_ca_dir contains no files, an empty JSON is generated and uploaded to the cluster, which removes all additional trusted CAs.
ocp_additional_trusted_ca_dir¤
Path to the directory containing the additional trusted CAs to configure in the cluster — it can contain none, one, or more CAs.
- Optional
- Environment Variable:
OCP_ADDITIONAL_TRUSTED_CA_DIR - Default:
/tmp
CA file names must match the registry URL and end with .crt (e.g. ibm.com.crt). A JSON file named additional_trusted_ca.json is created in the same path and deleted at the end. Only applies when ocp_update_additional_trusted_ca is true; currently works with ROSA HCP only.
Example Playbook¤
- hosts: localhost
any_errors_fatal: true
vars:
ocp_update_ciphers_for_semeru: True
ocp_ingress_update_timeouts: True
ocp_ingress_client_timeout: 30s
ocp_ingress_server_timeout: 30s
ocp_ingress_namespace_ownership: InterNamespaceAllowed
ocp_ingress_controller_name: default
ocp_operatorhub_disable_redhat_sources: True
ocp_enable_ipv6: True
roles:
- ibm.mas_devops.ocp_config
Setting Additional Trusted CA and Allowed Registries in Rosa HCP¤
- hosts: localhost
any_errors_fatal: true
vars:
cluster_type: "rosa-hcp"
cluster_name: "rosa-dev-cluster"
ocp_update_allowed_registries: True
ocp_allowed_registries:
- ibm.com
- mas.io
ocp_add_mas_registries: True
ocp_add_rh_registries: True
ocp_update_additional_trusted_ca: True
ocp_additional_trusted_ca_dir: /mnt/home/additional_trusted_ca
roles:
- ibm.mas_devops.ocp_config
License¤
EPL-2.0