Troubleshooting Oracle VM Database Crashes on ESXi with Guest-Level iSCSI Connections

Products

• VMware vSphere ESXi

Issue/Introduction

A Virtual Machine (VM) utilizing a direct Guest-level iSCSI connection reports database or block-level corruption

• The Oracle database stops unexpectedly when the VM is running on a specific ESXi host.
• Error Codes: DB errors oracle DB errors reported are ORA-00600, ORA-00448, or "Background Media Recovery terminated."
• The block Corruption may be reported with the following errors: 

Corrupt Block Found rdba: 0x00245b47
        TIME STAMP (GMT) = 03/24/2026 18:58:00
        CON_ID = 3, TSN = 6, TABLESPACE = PP110662_151122165046_DATA01
        AFN = 18, RFN = 1024, BLK = 2382663, RDBA = 2382663
        OBJN = 72051, OBJD = 72051, OBJECT = SYS_LOB0000072044C00006$$, SUBOBJECT =
        SEGMENT OWNER = CBSR, SEGMENT TYPE = Lob Segment
        CALLER = kcbi_check_fgio_main
        ENCRYPTION ERROR = 3 (BLOCK CHECK FAILURE)
        BLOCK CHECK ERROR = 6801 (LOGICAL)
        CORRUPTION ERR
Errors in file /local/rdbms/oracle/base/admin/diag/rdbms/pq######/PQ######/trace/PQ######_ora_######.trc  (incident=1461952) (PDBNAME=DB#####_###############):
ORA-01578: ORACLE data block corrupted (file # 18, block # 2382663)
ORA-01110: data file 18: '+DATA01/PQ######_########/ED######################/DATAFILE/pp#####_1_data01.286.1120842845'
PP110662_151122165046(3):Incident details in: /local/rdbms/oracle/base/admin/diag/rdbms/pq######/PQ######/trace/PQ######_ora_######.trc

 

Environment

vSphere 8.x

VCF 9.x 

Cause

ESXi does not have visibility into the SCSI payloads of in-guest iSCSI connections, fault isolation requires validating the networking and compute boundaries

Resolution

Perform the following checks to isolate the issue:

1. Validate ESXi Virtual Networking Configuration: Ensure the virtual machine's network adapter (vNIC) is connected to the correct port group. Verify the backing virtual switch (vSwitch or vSphere Distributed Switch) has active, healthy physical uplinks (vmnics) and that security policies are not inadvertently dropping traffic.

2. Analyze Network Traffic at the Hypervisor Level: Use the pktcap-uw utility on the ESXi host to capture packets at the virtual switch port and uplink levels during the issue occurrence. This verifies whether the encapsulated iSCSI frames are successfully entering and leaving the hypervisor. For exact syntax and usage, see Broadcom KB 341568.

3. Check Compute Resource Allocation: Review ESXi performance metrics (esxtop) for the affected virtual machine. High CPU Ready (%RDY) or Co-Stop (%CSTP) values indicate compute contention, which can starve the guest OS network stack and lead to dropped iSCSI connections. Ensure the virtual machine has adequate CPU/Memory limits and reservations configured in the vSphere Client. Refer: ESXTOP overview for Performance Troubleshooting