ecdye
commented
3 years ago First of all, thank you very much for zram-config. I have been using it with great results.
Thanks, I am glad you are enjoying it.
The thing I am unclear on, is I thought it was best practice to split up your zram devices by number of CPU cores.
I am not aware of anything that would make it more optimal to split up zram devices by the number of CPU cores, and as such it is probably unnecessary. Zram is fairly low impact on processing power anyway.
Note: I am only using zram for swap, not to store any files. But even if I create 4 lines in my ztab like:
# swap alg mem_limit disk_size swap_priority page-cluster swappiness swap zstd 512 2048M 75 0 90 swap zstd 512 2048M 75 0 90 swap zstd 512 2048M 75 0 90 swap zstd 512 2048M 75 0 90I end up with a single zram device of 2GB:
zramctl NAME ALGORITHM DISKSIZE DATA COMPR TOTAL STREAMS MOUNTPOINT /dev/zram0 zstd 2G 4K 70B 4K 4 [SWAP]
Well, I am not sure what would cause this behavior, it should still create all of the devices you request, although I believe it is unnecessary to split up zram devices by core, if you provide me with the debug log found at /usr/local/share/zram-config/log/zram-config.log I will attempt to debug and figure out why it might not be working.
I am also curious about your disk_size parameter, as zram-config is the only implimentation I see it in. Most other zram management utilities(including systemd-swap) set the disk size to the exact same size as the physical RAM that is being dedicated to zram. Doesnt that mean that the system will never try to pack more into the zram drive? Or does a zram device only report the compressed size of the data inside of it? And if that's the case, why would you want to specify a larger zram disk size than it's physical memory backing? Thanks!
The disk size is a different parameter because it represents the size of the files as they appear on the disk uncompressed. This is different because the compressed size is the limiting factor in RAM usage. The disk size should always be larger than the memory limit because it will be the uncompressed size you believe the directory will reach at maximum whereas the memory limit will be the compressed size that you believe the directory will reach at a maximum. The memory limit will change slightly depending on the efficiency of the compression algorithm being used.
I apologize for the delayed response, I have been very busy lately. I hope that I have helped to clarify what the disk_size parameter is meant for, please let me know of any further questions you may have.
stale[bot]
First of all, thank you very much for zram-config. I have been using it with great results. The thing I am unclear on, is I thought it was best practice to split up your zram devices by number of CPU cores. Note: I am only using zram for swap, not to store any files. But even if I create 4 lines in my ztab like:
I end up with a single zram device of 2GB:
I am also curious about your disk_size parameter, as zram-config is the only implimentation I see it in. Most other zram management utilities(including systemd-swap) set the disk size to the exact same size as the physical RAM that is being dedicated to zram. Doesnt that mean that the system will never try to pack more into the zram drive? Or does a zram device only report the compressed size of the data inside of it? And if that's the case, why would you want to specify a larger zram disk size than it's physical memory backing? Thanks!