3ware 9650SE SATA raid cards are a bit old, but I bet a lot of folks here are still using them somewhere, or did in the recent past ...
I have a 8 drive, raid6 array that lost one drive, and then another died during the rebuild and the replacement drive ended up being bad ... and then after replacing again, a third drive started to throw errors.
So, I set the card to ignoreECC during rebuild and just crossed my fingers. Things went well and the array is now back to fully healthy, BUT ...
... during the rebuild with no more parity left and a failing third drive, I got 21 of these errors:
c0 [Wed Oct 21 00:07:06 2015] ERROR Source drive ECC error overwritten: port=5
... which means that 21 times, data on my array got overwritten with just zeros because the card could not figure out the parity/checksum/math to reconstruct the bits there.
So I have data loss in 21 places in my array.
But how much ?
I cannot find any documentation or mention of the size of this data loss that this error represents. Does each error message represent a sector ? A block ? I know it doesn't represent something like a file, since the raid card has no idea what filesystem I am running on it.
Does anyone have any idea how much data each of these errors represents ?
(I do indeed have a backup, but if I can find out how much data these represent I can decide how much, if any, I want to bother to restore...)
Thanks.
I have a 8 drive, raid6 array that lost one drive, and then another died during the rebuild and the replacement drive ended up being bad ... and then after replacing again, a third drive started to throw errors.
So, I set the card to ignoreECC during rebuild and just crossed my fingers. Things went well and the array is now back to fully healthy, BUT ...
... during the rebuild with no more parity left and a failing third drive, I got 21 of these errors:
c0 [Wed Oct 21 00:07:06 2015] ERROR Source drive ECC error overwritten: port=5
... which means that 21 times, data on my array got overwritten with just zeros because the card could not figure out the parity/checksum/math to reconstruct the bits there.
So I have data loss in 21 places in my array.
But how much ?
I cannot find any documentation or mention of the size of this data loss that this error represents. Does each error message represent a sector ? A block ? I know it doesn't represent something like a file, since the raid card has no idea what filesystem I am running on it.
Does anyone have any idea how much data each of these errors represents ?
(I do indeed have a backup, but if I can find out how much data these represent I can decide how much, if any, I want to bother to restore...)
Thanks.