> Not sure what you mean, as I don't use mdadm... <br>This is the linux software raid utility.<br><br>> There is a BIOS level<br>> utility you access at boot time to add and remove drives from the array,<br>> which you can do with a failed drive after it has cooled off and started
<br>> working, if you only have one or two (after the first fails, the hot<br>> spare replaces it, when the second fails, the RAID5 still works but in<br>> degraded mode - it's when the third fails that problems set in, because
<br>> the other two have been marked bad and stay that way until you release<br>> them and reassign them).<br>This is the very similar with mdadm. But you with mdadm you can force the raid array to use failed drives without loosing their data. I assume you will have corruption at the points the array is out of sync but in your case if all drives are not really dead the array would have gone offline after 2 drives were kicked out so in that case the problem would have been recoverable if you figure out which drive was the second to be marked faulty.
<br><br>John<br>