When swap space is erased in /etc/init.d/halt.sh (by setting RC_SWAP_ERASE="yes" in /etc/conf.d/rc) swapon -a cannot mount swapspace on subsequent boots. Requires mke2fs [partition] and mkswap to fix. Found that out here http://bugs.centos.org/view.php?id=1399 Writing 0's to swap superblock not a great idea perhaps?
running mke2fs makes no sense at all the halt code runs `mkswap` on the partition after erasing it, so needing to run mkswap yourself also makes no sense post the output when shutting down