To make effective the changes you have to reboot if we are talking about your /.
To have compression if you install it on plain btrfs then you mount root sub volume with compression, an rsync things there.. While copying it will compress.
Lzo probably is best for compression but zlib is very lightweight on cpu.
Both will move ahead the write barrier.
Note: if you notice performance drop using discard you may disable it and clean the free space with fstrim from time to time. On sandforce controller I do like this.
Cheers!
Inviato da un telefono a gettoni con tapatalk
Installing LMDE on BTRFS with lzo compression, almost
Forum rules
LMDE 2 has reached end of support as of 1-1-2019
LMDE 2 has reached end of support as of 1-1-2019
Re: Installing LMDE on BTRFS with lzo compression, almost
The BTRFS wiki says the opposite about Zlib and Lzo:
https://btrfs.wiki.kernel.org/index.php/Compression
Lzo is designed to be real-time, compromising on compression ratio. But you said you tested it and found Zlib to perform better?
https://btrfs.wiki.kernel.org/index.php/Compression
Lzo is designed to be real-time, compromising on compression ratio. But you said you tested it and found Zlib to perform better?
R: Re: Installing LMDE on BTRFS with lzo compression, almost
By testing it seems that speed is almost same.keantoken wrote:The BTRFS wiki says the opposite about Zlib and Lzo:
https://btrfs.wiki.kernel.org/index.php/Compression
Lzo is designed to be real-time, compromising on compression ratio. But you said you tested it and found Zlib to perform better?
When i said zlib is better i was speaking about compression related to cpu stress.
But zlib is very light on cpu and very effectife in compression... may worth using it.
These are my 2c.
Inviato dal mio Transformer TF101G con Tapatalk 2
Re: Installing LMDE on BTRFS with lzo compression, almost
You could test which is better for, it depends on the CPU and HDD/SDD performance.
One uses lzo the other zlib (-3) so you could try to isolate compression by using lzop and gzip in a tmpfs and compare the result with your disk performance.
Would be nice to be able to select other compressions levels (-1 to -9) and have other compressors availabled (xz, lz4, etc).
One uses lzo the other zlib (-3) so you could try to isolate compression by using lzop and gzip in a tmpfs and compare the result with your disk performance.
Would be nice to be able to select other compressions levels (-1 to -9) and have other compressors availabled (xz, lz4, etc).