I’ve repeated the tests on my work laptop (i7) and I can’t seem to reproduce it, neither with the SATA SSD, nor with the NVME device.
It would be nice if a couple more people tried to test it once again, just to be able to close the bug if it is truly gone.
TY
Robert
P.S.
I use
vm.swappiness = 10
vm.vfs_cache_pressure = 50
Reconfigured from 32 GB memory / 0 GB swap to 8 GB / 16GB swap, with swappiness = 60 and vfs_cache_pressure = 100, presuming these to be the current defaults on Tumbleweed. Ran the following which would stall the machine immediately:
erlangen:~ # free -h
total used free shared buff/cache available
Mem: 5.6Gi 1.8Gi 1.7Gi 211Mi 2.2Gi 3.4Gi
Swap: 15Gi 0B 15Gi
erlangen:~ # dd if=/dev/sdb of=/dev/null bs=4M status=progress
500002979840 bytes (500 GB, 466 GiB) copied, 903 s, 554 MB/s
119235+1 records in
119235+1 records out
500107862016 bytes (500 GB, 466 GiB) copied, 903.185 s, 554 MB/s
erlangen:~ # free -h
total used free shared buff/cache available
Mem: 5.6Gi 2.2Gi 465Mi 235Mi 3.0Gi 2.9Gi
Swap: 15Gi 2.0Mi 15Gi
erlangen:~ #
With the dd running with foreground priority the machine stayed responsive. No annoyances encountered during testing.
Swapping indeed works. All applications stay usable while running the foreground dd:
karl@erlangen:~> free -h
total used free shared buff/cache available
Mem: 5,6Gi 4,1Gi 170Mi 599Mi 1,4Gi 681Mi
Swap: 15Gi 1,5Gi 14Gi
karl@erlangen:~>