Exadata Cloud Best Practices to Limit the Impact of Memory DIMM Failures (Doc ID 2815885.1)

Last updated on NOVEMBER 23, 2021

Applies to:

Purpose

When an Exadata Cloud DB Server experiences a full memory DIMM failure, it will likely crash, mark the DIMM as bad, and restart. X6, X7, or X8 systems do not reserve any memory to accommodate losing a DIMM and may be unable to allocate enough physical memory to restart after a DIMM failure. This note describes best practices to ensure a restart after a DIMM failure without manual intervention.

Scope

Applies to:
• Exadata Cloud@Customer X7, X8 (Gen 2) Quarter, Half, Full Racks
• Exadata Cloud@Customer X6, X7 (Gen 1) Quarter, Half, Full Racks
• Exadata Cloud@Customer X8, X8M, X9M Base Systems (Gen 2)

Note: This note does not apply to Exadata Cloud@Customer X8M and X9M Quarter, Half, and Full Rack systems. These systems have 1.5TB of memory, but only enable up to 1390GB to be allocated to the virtual machines, holding back enough that in the event of a DIMM failure, the full 1390GB can still be allocated to the VMs. So, with X8M and X9M, there is no need for a user to hold back any memory to ensure restart after a single DIMM failure. It also does not apply to Exadata Cloud@Customer X6 Eighth Racks, or X7 Base Systems, as these systems also have additional reserved memory.

Details

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.