You are not logged in.
Pages: 1

I've been doing some defrag operations on a USB HDD, and I'm seeing some weird usage patterns. I've looked into SMART status and I/O priority, but everything looks normal and changing I/O priority didn't do anything. Is this some kind of unavoidable throttling mechanism, or something I can fix? The filesystem is ext4 and I've enabled fast commits if that's relevant.
Offline
the graph doesn't make any sense - it shows the same what's written also as read? also: on the time scale and if the data size is somewhere accurate these are small bursts of small files - try a big 10GB file (or if the target has the space: 100GB to overcome cache) to see how it performs with bulk transfer in one go
Offline
The usage shown is from one big run of "sudo e4defrag -v /dev/sdb", so that's why the read and write graphs are mirrored. It's currently churning through a bunch of ~300MB mkv files. From the terminal output, I can see it stalls in the middle of processing individual files, it's not consistently stalling anywhere. Here are the results of transferring a large file: 
Copying from the HDD to my SSD is the 20:40-20:45 chunk and went very quickly. The rest is from copying from the SSD to the HDD. At the time shown, the file transfer was only about 50% done, but transfer rate was consistent. I have to assume the odd "spiking" behavior is unique to e4defrag.
EDIT: I realize I forgot to also screenshot the vertical axis, here's a better graph:
Last edited by wheelsbot7 (2026-01-05 14:37:18)
Offline
Pages: 1