I85PREP
⌘ K
CS AO study materials
CS AO study resources

CS AO: Storage and File System Administration

CS AO · Work through your exam, one topic at a time.

All guide sections

Storage and File System Administration

Storage and File System Administration

This domain covers capacity planning, RAID trade-offs, logical volume management, filesystem choice, and proving that data is where it should be.

Core rules

  • RAID: 0 stripes (no redundancy), 1 mirrors, 5 stripes with one parity disk (survives one failure), 6 uses two parity (survives two), 10 mirrors then stripes. RAID is not backup.
  • Partition tables: MBR is limited to 2 TiB and four primary partitions; GPT supports larger disks and many partitions. Use parted or sgdisk.
  • LVM layers: physical volume (PV) → volume group (VG) → logical volume (LV). Extend an LV and grow the filesystem online with lvextend -r.
  • Filesystems: ext4 is mature and supports online growth and offline shrinking; XFS excels at large files and parallel I/O but cannot shrink. NTFS/ReFS on Windows.
  • Monitor with df -h (capacity), df -i (inodes), du -sh, lsblk, and smartctl -a. Set quotas with xfs_quota or edquota.

Worked example

A /var volume is at 100% but du shows little data. Reasoning: check inodes first with df -i. A mail or session directory can create millions of tiny files and exhaust inodes while blocks remain free. If inodes are full, delete stale files (find /var/spool -type f -mtime +30 -delete) or rebuild with a higher inode count. If blocks are full, find large files with du -xh /var | sort -h | tail. Confirm a deleted file isn't held open by a process with lsof +L1.

Traps

  • Extending a partition but forgetting pvresize before lvextend.
  • Shrinking XFS - it is unsupported.
  • Assuming RAID protects against deletion, ransomware, or controller failure.
  • Ignoring inode exhaustion because du looks normal.