Skip to content

RAC nodes fail with network heartbeat loss (gipcretConnectionLost) on Oracle Linux 8 (GCP VM, Podman environment) #3036

@akshay-das

Description

@akshay-das

Description

I’m following the RAC container setup documentation using the setup_rac_host.sh script to spin up RAC containers via Podman.

Environment details:

  • Platform: Google Cloud Platform (GCP) VM
  • OS: Oracle Linux 8

Reference: OracleDatabase/RAC/OracleRealApplicationClusters

The script successfully provisions the RAC nodes (racnodep1, racnodep2), but during the cluster startup phase, I encounter repeated gipcretConnectionLost and no network HB messages, indicating the interconnect heartbeat is failing.

Observed Logs

Below is a snippet of the recurring error messages from racnodep2:

2025-10-24 17:46:42.320 :    ONMD: [INFO] clssnmvDHBValidateNCopy: node 1, racnodep1, has a disk HB, but no network HB
2025-10-24 17:46:42.391 :GIPCHAUP: [INFO] gipchaUpperProcessDisconnect: processing DISCONNECT for hendp 0x7f140c03e570
2025-10-24 17:46:42.391 :GIPCGMOD: [INFO] gipcmodGipcCallbackDisconnect: Disconnect forced for endp 0x7f140c042c10
2025-10-24 17:46:43.391 :    ONMD: [INFO] clssnmDiscHelper: racnodep1 connection failed, endp (0x9d7)

Expected Behavior

RAC containers should establish stable interconnect communication (gipcha and clssnm layers) between nodes when configured through setup_rac_host.sh.

Request

Could the maintainers please:

  • Confirm if RAC-on-Podman is supported on Oracle Linux 8 under GCP?
  • Suggest any required network adjustments (bridge, MACVLAN, multicast) for the interconnect?

Metadata

Metadata

Assignees

Labels

RACOracle Database RAC

Type

No type
No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions