Summary
Kernel logs a warning message in syslog and hangs while or after applying GI RU 19.19.0.0.230418 (April'23).
Symptoms
When Oracle tries to load the oracleoks module, the following messages will be logged in syslog:
Apr 20 08:14:00 rac1 kernel: watchdog: BUG: soft lockup - CPU#3 stuck for 41s! [modprobe:53251]
The system will become unresponsive after a while.
Affected Products
FlashGrid Cluster on AWS/Azure/GCP
Affected Versions
RHEL 8.7 (or Oracle Linux 8.7 with RHCK kernel) kernel versions: 4.18.0-425.10.1.el8 or higher.
Grid Infrastructure April 2023 Release Update 19.19.0.0.230418 (or any Grid Infrastructure version with patch 34718455 installed).
RHEL 8.8 kernels 4.18.0-477.10.1.el8_8 and above are not affected.
Root Cause
oracleoks module included in GI RU 19.19.0.0.230418 (and also in patch 34718455) is not compatible with RHEL 8.7 kernel versions 4.18.0-425.y.z.el8_7 However, it attempts to load causing the soft lockup.
Resolution
RHEL 8.7: If you want to apply the release update, install the Grid Infrastructure April 2023 Release Update 19.19.0.0.230418 only in combination with GI patch 35068505 and in manual (no opatchauto) mode.
If you want to update the kernel to one of RHEL 8.7 kernels 4.18.0-425.y.z: install GI patch 35068505 before updating the kernel.
If ACFS is not used: update to RHEL 8.8 kernel 4.18.0-477.10.1.el8_8 or higher, and apply the Grid Infrastructure April 2023 Release Update 19.19.0.0.340518 (no need to install patch 35068505).