fsck is probably wrong in guessing that it’s a zero-length partition; I would be more-inclined to guess that a read error caused it not to be able to read anything.

# parted /dev/sdd print

That should display the partition table on the drive, as well as the lengths of the partition. My guess is that it’ll just show that you have a 4.5TB partition. Probably starts at something like 1MB and ends at something like 4500GB.

Linux was able to read from the drive enough to get at the partition table on it, or the lsblk output you had above in that image wouldn’t be able to show the partition.

But I’d guess that it’s hitting read errors in trying to read from the drive, and that the way that this is percolating up to fsck is fsck asking for N bytes and getting back only the bytes that it could read before it hit the error.

It’s maybe not impossible that an invalid partition table could produce that error. And maybe e2fsck is trying to read off the end of the device or something because the filesystem on the partition is messed up, but I’m biased towards to thinking that it’s more-likely that the drive is failing.

The first option I’m normally going to take when maybe a hard drive is dying is:

# smartctl -a /dev/sdd

As long as the drive supports SMART, that’ll tell you whether the drive thinks that it’s failing. That won’t catch everything, but if the drive doesn’t give itself a clean bill of health, then I wouldn’t either.

If the kernel is trying to read data from a drive and seeing errors, it’ll probably show up in the kernel log.

# dmesg | grep sdd

Should show you errors in the kernel log from the current boot relating to that drive.

If you don’t see any errors there, then I’d be inclined to see whether it’s possible to read the contents of the partition and whether that can do so. Something like – and be very careful not to reverse the “if” (input file) and “of” (output file) parameters here, because dd can write to a drive and if typed incorrectly, this can overwrite your drive’s contents:

# dd if=/dev/sdd1 status=progress of=/dev/null

That’ll just try reading from the beginning of the partition, giving you updates as it goes to let you know how much it’s read, and then just dump the data into /dev/null. If it fails to read data at some point, it should bail out way before 4.5TB. Probably don’t want to wait for the whole partition to be read, though, because 4.5TB is gonna take a while.

If dd can read the contents of the partition, then I assume that e2fsck should be able to as well.

If dd can read the data, then I’d assume that something is wrong with the filesystem, or at least the data on the partition, and it’d make sense to look at that.

If dd can’t read the data, then I’d be inclined to blame hardware (or maybe the partition table), since nothing inside the partition itself, including the filesystem, should be able to make dd fail to read the contents of the partition; dd doesn’t actually pay any attention to the data that it’s reading.