Skip to content

KB-3-Storage.md

Latest commit

 

History

History
192 lines (146 loc) · 9.17 KB

KB-3-Storage.md

File metadata and controls

192 lines (146 loc) · 9.17 KB
Raw

Storage Managment

linux device handling

  • the linux kernel is responsible for device managment
  • it uses modules to work with specific device types
  • kernel modules are included in the initramfs, or can be loaded on demand
  • initramfs (initrd in some distos) is compiled while installing linux
  • automatic on-demend device initialization (PnP) is done by systemd-udevd
  • manual device loading can be done using modprobe
  • to interface with devices, most (but not all) devices have a device node in /dev
  • hardware related params are stored in /sys and its subdirs

listing devices utils

lsdev - information about installed hardware
lsusb - list usb information
lspci - list pci connected device information
lsblk - list block devices
lscpu - display CPU information
dmesg - print or control the kernel-ring buffer

systemd udevd

  • enables plug and play behavior
  • when alerted by the kernel systemd-udevd will initialize new devices
  • this involves writing device info to /sys and creating device nodes in /dev
  • udevadm monitor to trace the procedure of device initialization
  • udevadm info --name=/dev/<node> to get detailed information about devices

pseudo filesystems

  • the kernel uses pseudo filesystems - meaning they exist in memory only and not on disk
    • /proc is for kernel runtime information and tunables
    • /sys is for hardware related information
    • /dev is used to create device nodes

managing partitions

  • Master Boot Record (MBR) is the older partition kind

    • works wiith BIOS to store information in a 512kb on-disk area, as well as the boostrap loader
    • 4 primary partitions are supported
    • to address more partition, extended and logical partitions are needed
    • a maximum of 2TB can be addressed
    • fdisk is the recommended utility to create MBR partitions

  • Guid Partition table (GPT) is the newer partition kind used to address the shortcomings of MBR

    • larger are to store up to 128 partitions
    • no need for extended partitions
    • Required on systems that work with UEFI
    • parted and gdisk are the tools to work with GPT partitionsא

swap partitions

  • swap is used to cache inactive application memory

  • more memory available for caching files and other information on behalf of RAM

  • if only inactive application memory is swapped, performance won't be effected

  • linux servers should have a minimum of swap to alllow for swapping out all inactive application memory

  • as it works on contiguous disk space, using a swap device is recomended

  • mkswap - set up a swap area

  • swapon/swapoff - enable/disable devices and files for paging and swapping

  • free - Display amount of free and used memory in the system

file systems

  • there are many file systems available on linux

    • Ext3 - journaling file system used by Linux that provides better reliability and data recovery than its predecessor, Ext2
    • Ext4 - successor to Ext3 that adds support for larger file sizes and improved performance, while maintaining compatibility with Ext3
    • XFS - high-performance journaling file system originally developed by Silicon Graphics for their IRIX operating system, now commonly used on Linux servers and workstations
    • NFS - network file system protocol used for sharing files over a network between UNIX-based systems
    • SMB/CIFS - network file system protocol used for sharing files over a network between Windows-based systems
    • FAT - network file system protocol used for sharing files over a network between Windows-based systems
    • NTFS - proprietary file system used by modern versions of Windows that supports large file sizes and advanced security features
    • Btrfs - modern copy-on-write file system designed for Linux that provides features such as snapshots, checksums, and RAID-like functionality

  • to use a partition you need to put a filesystem on top

  • different file systems exist, but ext4 and xfs are the defaults

  • use mkfs.<fs-kind> to create a file system

  • you can use -L to give the file system a label

  • use mount to connect the file system to a directory as its mount point

automating mounts

  • /etc/fstab is used to automate mounts on between boots
  • it is used as an input file for fstab-generator, a systemd component that generates systemd mounts
  • in /etc/fstab, 6 columns are used to automate mounts, last 2 are deprecated
    • device
    • mount point
    • file system type
    • mount options
    • dump option (DEPR)
    • fsck option (DEPR)

potential issues

  • block device names may changed down the road (i.e. extended MBR partition numbers )
  • if you're refering to a block device name in /etc/fstab, that can create problems
  • avoid these problems using UUIDS and labels
  • when creating a file system, UUIDS are automatically assigned
  • use blkid to get the uuid for your filesystems, labels are assigend manually
  • to add label after filesystem creation use a specific utility per fs
    • xfs_admin -L for XFS
    • tune2fs -L for Ext4
  • it is useful to check /etc/fstab if any device issues are occuring to validate presistent mount between boots

LVM

  • Provides a layer of abstraction between physical storage devices and file systems
  • Allows creation of virtual storage volumes that span multiple physical disks
  • Enables resizing of volumes on-the-fly without data loss
  • Facilitates moving data between physical devices without downtime or data loss
  • Can merge logical volumes together for greater flexibility in managing storage resources
  • Provides high availability and scalability in enterprise environments

To use LVM in Linux, follow these basic steps:

    1. Create one or more physical volumes (PVs) by partitioning your storage devices or disks using a tool like fdisk or gdisk.

      • you can set partition type to be 8e or 8e00 for which is a low level identifier, it's considered best practice

    2. Add your physical volumes to a volume group (VG) using the pvcreate and vgcreate commands.

      • pvcreate /dev/nnn to mark partition as physical volume
      • vgcreate vgname /dev/nnn to create a VG that contains at least 1 partition

    3. Create one or more logical volumes (LVs) within the volume group using the lvcreate command.

      • lvcreate -L <size>G -n lvname vgname

    4. use pvs, vgs, lvs or pvdisplay, vgdisplay, lvdisplay to varify

    5. Format the logical volumes with the desired file system using a tool like mkfs.

    6. Mount the logical volumes to a mount point of your choice using the mount command.

  • To resize an existing logical volume, use the lvresize command.

  • To move data between physical devices, use the pvmove command.

  • To merge logical volumes, use the lvmerge command.

  • To remove a logical volume or a volume group, use the lvremove or vgremove commands, respectively.

  • To remove a physical volume, use the pvremove command.

RAID configurations

  • Provides data redundancy and/or increased read/write performance by combining multiple physical disks into a single logical unit
  • Allows hot-swapping of failed disks without downtime or data loss
  • Supports various RAID levels, each with its own trade-offs in terms of performance, redundancy, and capacity utilization
  • Can be implemented at the software or hardware level, hardware RAID is more advised
  • different RAID types are offered
    • RAID 0: striping
    • RAID 1: mirroring
    • RAID 5: striping with distributed parity
    • RAID 6: striping with dual distributed parity
    • RAID 10: mirrored and striped
  • To create a software RAID configuration in Linux, follow these basic steps:
    • Partition the physical disks you want to use for RAID using a tool like fdisk or gdisk.
      • fdisk /dev/sdb; fdisk /dev/sdc; use type RAID
    • Create a new RAID array using the 'mdadm' command, specifying the RAID level, the disks to include, and any other relevant options.
      • mdadm --create /dev/md0 --level=<raid type> --daid-disks=2 /dev/sdv /dev/sdc (2 disks in this case)
    • Create a file system on the new RAID array using a tool like mkfs.
      • mkfs.ext3 /dev/md0
    • Mount the new RAID array to a mount point of your choice using the 'mount' command.
      • mkdir /raid
      • mount /dev/md0 /raid
    • To replace a failed disk, hot-swap the disk with a new one of the same size and type, then use the 'mdadm' command to add the new disk to the RAID array and rebuild the data.

monitoring filesystems and disk usage

disk usage

df - shows the amount of space available on a mounted device
du - shows disk usage per directory

file systems

tune2fs - show and optimiz Ext file systems dumpe2fs - dump Ext file system metadata xfs_admin - monitor XFS properties

troubleshooting

fsck - monitor file system integrity
partprobe - update kernel partition table with current storage information
du - show overview of available disk space per directory
df - show available disk space per device
iostat - storage related statistics
ioping - used to test IO performance

  • to optimize IO performance, the IO scheduler can be tuned through /sys