Additional Information for Troubleshooting Memory or Battery Errors on the PERC Controller

A RAID Controller error message is displayed during POST to indicate that the controller's cache does not contain all of the expected information, or it contains data destined for a hard drive that cannot be or has not been written to the drive. The most common reasons why this error may occur are:

• Server did not perform a normal shutdown process – Power loss and/or spontaneous restarts can result in incomplete or corrupted data to remain in cache that cannot be written to a drive.

• Cache memory is defective – Bad cache memory can cause data to become corrupted. This can cause OS-related issues and spontaneous reboots.

• Loss of battery power while server is shutdown – Controllers that do not use NVCACHE (Non-Volatile Cache) memory utilize batteries that can retain the contents of cache for a limited time (24-72 hours) while the server is not powered on. Once the battery drains, the entire contents of cache is lost and the controller recognizes that the cache memory does not contain all of the information expected. Controllers that do utilize NVCache (some H700/H800 controllers and newer controllers such as H710, H710P, H810) are very unlikely to encounter this issue since the battery only needs to maintain power for 30 seconds or less in most cases.

• PERC Battery Maintenance

A PERC battery that is suspected to have failed or has a warning symbol displayed in OpenManage Server Administrator should have a manual Learn Cycle performed. A Learn Cycle causes the battery to discharge and recharge, and restores the battery to a fully functional condition. In some cases, multiple Learn Cycle procedures may be required to restore the battery to an effectively charged state. To perform a manual Learn Cycle, select Start Learn Cycle from the Battery Tasks drop-down menu in Open Manage Server Administrator (OMSA).

• Cache Use

Hardware RAID controllers utilize cache (a temporary repository of information) for their normal operation. The normal operation cache comprises DRAM memory, which, like system memory, retains data only when powered on.

Newer controllers utilize NVCache, which is utilized when the server is powered off. NVCache memory contains both DRAM memory (for normal operation) and flash memory (non-volatile). The controllers battery (if operational) powers the DRAM memory during a power loss so that the contents can be copied into the flash memory for indefinite storage.

The contents of cache can essentially be broken into three parts:

• RAID configuration and metadata - Information about the RAID arrays including configuration information, disk members, role of disks, etc.

• Controller logs - RAID controllers maintain several log files. Dell technicians rely on the TTY log as the primary log for troubleshooting various RAID and hard drive issues.

• RAID data - This is the actual data destined to be written to the individual hard drives. Data is written into the cache of the controller in both Write Through and Write Back cache policy modes.