Manage disk partitions

I have a replacement disk to install this afternoon. Create a new partition table on /dev/sdb with a single primary ext4 partition using the whole disk, set it bootable if the host uses BIOS, format it, add it to LVM as pv if requested, update /etc/crypttab and /etc/fstab if the disk holds encrypted volumes, and verify the device shows correct size and filesystem UUIDs before handing back to ops. Do not touch /dev/sda.

Before I partition multiple machines, make the layout reviewer-friendly. Show the partition sizes and labels you propose for /dev/sdb, explain why each partition exists (boot, root, swap, data), surface any alignment or RAID considerations, and flag anything that could break existing backups or monitoring. Recommend the one change that minimizes risk during mass rollout.

While prepping servers for a storage upgrade I mistakenly wrote a new partition table to the production boot disk instead of the spare. The machine still boots for now, but I fear I’ve corrupted metadata and could lose the install on the next reboot. I don’t know if the previous table is recoverable or if I should power off and image the disk. What is the most likely extent of damage, what reads can I do safely to assess recoverability, and what immediate actions preserve the option to recover the old table?

We keep having one-off partitioning errors during maintenance windows that end up requiring restores and long outages. I want fewer human slips. Which three process changes or tools would most reduce accidental writes to wrong disks — such as standard device naming, preflashing disk labels, or mandatory peer checks — and which should I implement first to get the most protection fast?