https://www.maketecheasier.com/move-home-folder-ubuntu/
To set last boot OS as default boot, courtesy of AskUbuntu:
GRUB_DEFAULT=saved GRUB_SAVEDEFAULT=true
sudo update-grub
In the (un)fortunate event you need to create a new USB stick, can follow the steps below, referenced from AskUbuntu:
# Insert USB stick lsblk # check location of disk and filesystem, alternatively with 'df' sudo umount /dev/sda # unmount disk/filesystem # Write to flash stick with dd utility sudo dd bs=4M if={{PATH_TO_ISOFILE}} of={{PATH_TO_DISK}} conv=fdatasync status=progress && sync 3825205248 bytes (3,8 GB, 3,6 GiB) copied, 319 s, 12,0 MB/s 912+1 records in 912+1 records out 3826831360 bytes (3,8 GB, 3,6 GiB) copied, 937,909 s, 4,1 MB/s
If just reinitializing the partitions, overwriting the MBR section is sufficient:
dd if=/dev/zero of=/dev/sdX bs=512 count=1 seek=0
To eject or mount a USB:
# No umount necessary, will be handled by eject sudo eject /dev/sdX sudo eject -t /dev/sdX
fdisk
can be used.
# List disks and partitions user:~$ sudo fdisk -l # Select unallocated disk, e.g. /dev/sda user:~$ sudo fdisk /dev/sda Command (m for help): n # for new partition Partition number (1-128, default 1): First sector (...): Last sector (...): Created a new partition 1 of type 'Linux filesystem' and of size ... # Commit changes Command (m for help): w
This was confusing to get into, so some guides to start out:
An executive summary:
Some useful commands:
root:~# mkfs.btrfs -L mylabel /dev/sda
First verify if it is really due to BTRFS, e.g. space could be consumed by excessively verbose logs. The usual du
techniques work. Some other tools:
du --max-depth=1 -h .
summarises the available disk space in the current directorydf -i /
checks if there are available inodeslsof | grep deleted
checks if any processes are still holding on to space allocated to already deleted files
Check the BTRFS version with btrfs version
. Some important information:
sysfs
knob
To see how much space BTRFS is consuming, use the btrfs
program to check: fi show
for device overview, fi df /
for data/metadata breakdown, and fi usage /
for a more detailed breakdown across devices.
# Or use the shortcut 'btrfs fi show' root:~# btrfs filesystem show Label: none uuid: 080a6971-d262-42cb-af4d-47571f31a03a Total devices 1 FS bytes used 15.79GiB devid 1 size 26.63GiB used 26.63GiB path /dev/mmcblk0p2 root:~# btrfs fi df / Data, single: total=24.61GiB, used=14.06GiB <<-- BAD System, single: total=32.00MiB, used=16.00KiB Metadata, single: total=1.29GiB, used=1.21GiB GlobalReserve, single: total=67.75MiB, used=0.00B
The failure mechanism above is due to the inability to allocate more chunks for the metadata block group. This only occurs under the low disk space condition (see UnixSE). Under normal conditions, the metadata group will be automatically allocated more chunks as needed using unallocated disk space.
Balancing of the filesystem is needed, to reallocate data out of chunks to free them up for use. A "null rebalance" may be sufficient, which triggers deletion of empty chunks:
root:~# btrfs balance start -dusage=0 -musage=0 / & root:~# btrfs balance status -v # check status of background job root:~# btrfs balance cancel / # cancel balancing background job
If null balancing reports that no data was relocated any chunk, then slowly increase the -dusage
and -musage
parameter to increasingly clear out chunks with allocations lower than the specified percentage:
root:~# btrfs balance start -dusage=1 / Done, had to relocate 0 out of 39 chunks root:~# btrfs balance start -dusage=5 / Done, had to relocate 1 out of 39 chunks root:~# btrfs balance start -dusage=10 / Done, had to relocate 0 out of 38 chunks root:~# btrfs balance start -dusage=20 / Done, had to relocate 3 out of 38 chunks root:~# btrfs balance start -dusage=40 / Done, had to relocate 4 out of 37 chunks
Consider doing this for the metadata group as well (using -musage
), though note that . Guides may suggest the use of parameters -dlimit
and -mlimit
, which limits the number of reallocations performed in the current balancing run. Not particularly useful when clearing up as much space as possible, but useful for cron scripts, e.g.
24 */1 * * * btrfs balance start -dusage=20,limit=5 / >>/tmp/btrfs_balance.log 2>&1
Note also the existence of scrub
command, but this is more for correcting errors across RAID volumes, rather than a space issue.
In other words, an ENOSPC
error. If ENOSPC is triggered by the balancing itself, one needs to clear emergency space. Some available options:
truncate -s0 FILE
. This reduces space without requiring extra metadata.To clear emergency space, start by deleting old BTRFS snapshots.
root:~# btrfs sub show / @/.snapshots/1/snapshot Name: snapshot Snapshot(s): @/.snapshots/24/snapshot @/.snapshots/25/snapshot root:~# btrfs fi du -s /.snapshots/* Total Exclusive Set shared Filename 6.78GiB 0.00B 6.78GiB /.snapshots/1 6.78GiB 0.00B 6.78GiB /.snapshots/24 6.78GiB 0.00B 6.78GiB /.snapshots/25 root:~# btrfs subvolume delete /.snapshots/24
If a USB stick or hard drive is available on hand, add the device to BTRFS - this increases the amount of available space on the system. Find the device using lsblk
.
root:~# btrfs device add /dev/sda root:~# btrfs filesystem show / # verify two devices available root:~# btrfs filesystem df / # check available unallocated space
If this device is only a temporary one for running balance
, then perform a small rebalance first, before removing it and doing the full balance:
root:~# btrfs balance start -dusage=5 / root:~# btrfs device remove /dev/sda / root:~# btrfs balance start -dusage=40 / # rebalance without the additional device