Skip to content

Latest commit

 

History

History
93 lines (76 loc) · 4.19 KB

NEW_DRIVE.md

File metadata and controls

93 lines (76 loc) · 4.19 KB

Add a new Disk

Install the new disk

  • Shutdown the server safely
  • Remove any disks that need to be removed
  • Install any new disks
  • Start up the server

Determine the disk information

  • Determine what disks are currently mounted
    • df -h
  • Determine the device id's for new disks
    • ls -l /dev/disk/by-id
    • ls -l /dev/disk/by-label
  • Verify the device id's match your new disks using disk info
    • smartctl -a /dev/sdX - replace sdX with your device id.

(Optional) Verify the disk health

  • This step can take hours, and is not really required as Mediadepot has a cronjob that tests for bad blocks and runs S.M.A.R.T tests continuously.
  • Run badblocks
    • badblocks -c 65536 -b 4096 -vs /dev/sdX

Format the new disk

  • For your new disks, format them with ext4
    • mkfs.ext4 -L "8.0TB-WD-3" -j /dev/sdX - replace 8.0TB-WD-3 with disk size, manufacturer and slot#. replace sdX with device id.
    • When prompted, proceed to format the entire device using 'y'
  • Tell the kernel to re-read the partition table
    • sfdisk -R

Mount the disk on startup

  • Identify the hard drive to mount and copy the UUID from the output that corresponds to the hard drive (i.e. sdb)
    • ls -l /dev/disk/by-uuid/
  • Add the UUID to the /etc/fstab, replacing the UUID as captured in the previous step
    • nano /etc/fstab
  • If this is a new disk, make sure you also add it's mount location to the MergerFS line.
#
# /etc/fstab
# Created by anaconda on Sat Nov  9 01:46:39 2013
#
# Accessible filesystems, by reference, are maintained under '/dev/disk'
# See man pages fstab(5), findfs(8), mount(8) and/or blkid(8) for more info
#
UUID=1ebbf241-528c-465e-889f-acc15400dd8c /                       ext4    defaults        1 1
UUID=75346b8e-b162-458c-b0e9-a8d48ec2bc82 swap                    swap    defaults        0 0
UUID=ad85eeb9-18f0-4b85-9bfa-b88a5d1489b3 /mnt/drive1             ext4    defaults        0 2
UUID=547b073d-e591-4913-b4fb-7c5084353979 /mnt/drive2             ext4    defaults        0 2

/mnt/drive1:/mnt/drive2:/mnt/drive3 /media/storage fuse.mergerfs direct_io,defaults,allow_other,minfreespace=50G,fsname=mergerfs 0 2
  • Test the modified fstab file
    • mount -a - If all goes well, there should not be any output. If there are errors, stop and diagnose the problem.
    • a response of fuse: mountpoint is not empty can be ignored. its a message from MergerFS.

Add the disk to MergerFS

  • Temporarily add the new disk mount path to the MergerFS system.
    • xattr -w user.mergerfs.srcmounts +/mnt/driveX /media/storage/.mergerfs - replace driveX with mount location
  • Rebalance the MergerFS disks
    • mergerfs.balance /media/storage

(Optional) RSync over any data from previous disk.

This command can be used to synchronize a folder, and also resume copying when it's aborted half way. The command to copy one disk is:

rsync -avxHW --numeric-ids --info=progress2 /mnt/disk_source/ /mnt/disk3/

The options are:

-a  : all files, with permissions, etc..
-v  : verbose, mention files
-x  : stay on one file system
-H  : preserve hard links (not included with -a)
-W  : copy files whole (without rsync algorithm)
--remove-source-files : delete source files on sync complete

# The following options are not supported in CoreOS/Flatcar
-A  : preserve ACLs/permissions (not included with -a)
-X  : preserve extended attributes (not included with -a)
--numeric-ids : don't map uid/gid values by user/group name
--info=progress2 : instead of --progress is useful for large transfers, as it gives overall progress

Trailing slash on /mnt/disk_source/ is important as it means that the content of disk_source folder will be copied into the destionation folder, instead of copying the disk_source folder into the destination.

Resources