Some amount of data loss seems to be the expected norm with Write Back. This Write Policy seems to cause file system corruption in FxM appliances that gradually builds up over a period of weeks and months. Eventually, the filesystem corruption becomes so severe that the appliance will no longer be able to function. That is why we recommend rebooting every 90 days, before the filesystem corruption becomes to severe for the automatic fsck (for "file system consistency check") to repair. Doing this helps but is not 100% effective.
Usually the extent of the filesystem corruption that has been latently building up suddenly manifests upon reboot, and the appliance will not be able to come back up successfully. At this point, the user needs to reboot from the Rescue DVD to run and run fsck manually from the unmounted filesystem, to fix the corruption, as per this Solution article: https://support.quest.com/foglight-experience-monitor/kb/54010
The write policy on the R710 and R720 PERC controllers can be set to either "Write Back" or "Write Through". When set to "Write Through", the controller sends a data transfer complete signal to the OS when the disk subsystem has received all the data. When set to "Write Back", the controller sends a data transfer complete signal to the OS when the controller cache has received all the data. When "Write Back" mode is enabled, the controller will flush the data out of the cache in the background. This significantly speeds up the disk write performance.
So how do we configure the Foglight Experience Monitor (FxM) RAID controller 'Write Policy to be 'Write Through'? (as opposed to the current default setting of 'Write Back') on a Dell R710/R720/R30
Data loss seems to be the expected norm when the Write Policy is set to Write Back. FxM cannot tolerate any type of data loss events. This is being tracked as FXM-685.
NOTE: Changing the Write Policy to Write Through is not 100% effective in preventing this problem from occurring but it greatly reduces the likelihood.
NOTE: Please refer to the Word document in the 'Attachments' section named "RAIDWriteThruConfig" for screenshots of each step. The instructions immediately below are just the text of the steps and so not as fully illustrative.
1. Reboot the machine
2. During the startup sequence, you will see the following. Press Ctrl-R to start the Configuration Utility when this is displayed
PowerEdge Expandable RAID Controller BIOS
Press <Ctrl><R> to Run Configuration Utility
3. Once the Configuration Utility has started, press F2 for the Operations menu
4. Select Properties from the Operations Menu
5. On the Virtual Disk Management dialog, Tab to the Advanced Settings button and press the <Space Bar> to select
6. Tab to the Write Policy drop down and press <Enter>
7. Select "Write Through" and press <Enter>
8. Tab to the OK button and press <Enter>9. Hit <Esc> to exit the Configuration Utility
10. A prompt will ask "Are you sure you want to exit?". Select OK and press <Enter>