This content has been marked as final. Show 6 replies
Hi i found this today, so i will have to replace a new Battery before furthur tracking.
A battery fault can cause write cache to become disabled. Write cache will not be re-enabled until the battery has been replaced and fully charged. A newly replaced battery can take upwards of 12 hours to recharge.
Pasted from <https://support.oracle.com/epmos/faces/ui/km/SearchDocDisplay.jspx?_afrLoop=1483553035068416&type=DOCUMENT&id=1411763.1&displayIndex=2&_afrWindowMode=0&_adf.ctrl-state=106iryozc1_173>
Thanks and Regards.
Battery is there to ensure write cache can be maintained during a power failure. Enabling write cache without battery (btw, not a user option) is a dangerous proposition bec. once an I/O request is written to cache by the controller, it is considered complete and the host application will be informed with an i/o success. The lost of write cache before the content is written to disk, means data lost that is completely undetectable by the host application unless the host file system structure provide alternative error detection mechanism (such as a block level checksum). This data corruption can potentially go undetected forever since as far as the host is concerned it was successfully written, while on the storage side, the RAID structure (if any) is intact and there would be no parity check error since write cache was "lost" and not bec. it was partially written.
It's a good idea to keep write cache disabled while waiting for a new battery.
It can be done. If you have SANtricity, it's an option you can easily select.
With CAM, you have to use the CLI. See:
Page 177 or thereabouts is apropos.
But as pointed out, doing so is playing with serious fire. You shouldn't even think about doing it unless your array is on a UPS that you trust entirely.