e2fsck 1.41.14 (22-Dec-2010)
The filesystem size (according to the superblock) is 244061472 blocks
The physical size of the device is 244061454 blocks
Either the superblock or the partition table is likely to be corrupt!
/dev/md0 contains a file system with errors check forced.
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
/lost+found not found. Create? no
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/md0: ********** WARNING: Filesystem still has errors **********
/dev/md0: 180825/61022208 files (12.2% non-contiguous) 240908460/244061472 blocks
raid1: raid set md0 active with 1 out of 2 mirrors
md0: detected capacity change from 0 to 999675715584
md0: unknown partition table
EXT4-fs (md0): bad geometry: block count 244061472 exceeds size of device (244061454 blocks)
If you have read the previous posts, you know that the e2fsck warning is just a symptom of the problem. That's why the resize2fs (and e2fsck) is run on the partition, and not on the array.
More than 98% used is indeed a problem. You could use the -M flag (man resize2fs), but I don't know how 'good' resize2f is. It won't kill the filesystem, but I don't know if it will squeeze enough files in unused space to be able to shrink the partition. On the other hand, you only need to shrink it 18kB.