H205534: Multi-Bit Error Correcting Code (ECC) will cause incorrect data to be written

Symptom

When an IBM System Storage DS3500 Storage Subsystem controller or DCS3700 Storage Subsystem controller has detected a Multi-bit Error Correcting Code (ECC) memory error, users will see an informational message in the Major Events Log (event 0x2604) and that individual controller will reboot.

DS3500 systems running controller firmware versions 7.70.45.00 or older, 7.75.11.00, or 7.77.29.00 or older and DCS3700 systems running 7.77.29.00 or older may not handle this Multi-bit ECC memory error properly, which will result in incorrect data being written.

Affected configurations

The system may be any of the following IBM servers:

• IBM System Storage DCS3700 Storage Subsystem, type 1818, any model
• IBM System Storage DS3512, type 1746, any model
• IBM System Storage DS3524, type 1746, any model

This tip is not option specific.

Solution

This behavior has been corrected in the DS3500 and DCS3700 controller firmware version 7.77.34.00.

These firmware updates are available by selecting the appropriate Product Group, System Storage type, Product name, and operating system on IBM Support's Fix Central web page, at the following URL:

• http://www.ibm.com/support/fixcentral
• IBM highly recommends that users upgrade all of their DS3500 and DCS3700 controllers to 7.77.34.00 firmware version immediately.
  

A DS3500 and DCS3700 controller firmware issue caused this behavior. It has been corrected as described above.

Note: When a Multi-bit ECC memory error occurs while running firmware version 7.77.34.00, the 0x2604 MEL event will still be logged and the controller will still reboot, but the error recovery will be handled properly.

This issue has not been reported to IBM by any IBM user.