I guess this is what I get for thinking that chunk size didn't matter for RAID1 arrays when I set up a few of them a couple years back on my Alphaserver running Debian Linux... And then choosing or defaulting (don't recall which) an absurdly small value. :-) A recent attempt at switching from my 2.6.15 kernel to a newer 2.6.16 kernel resulted in the above error when attempting to assemble /dev/md1, which is my root filesystem. Looking at the source trees for 2.6.15 and 2.6.16 shows the test hasn't changed, so I conclude that my immediate problem is that the new kernel has decided I would be better off with 8KB pages rather than 4KB pages. Whatever. Since I'm using RAID1, for which chunk_size reputedly is a meaningless quantity, what I'd really like to do is change that on my existing arrays to something else (64KB?) and just carry on. However, the mdadm v1.12.0 I have from Debian provides no mechanism for doing that, and I don't recall discussion of this functionality being added to any of the newer mdadm releases either. Does anybody have a clever suggestion, or do I grit my teeth and CAREFULLY recreate my arrays while trying not to destroy any of the encapsulated data? Thanks, Scott Bailey scott.bailey@xxxxxxx - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
