I’ve installed SeaDrive on a Linux (Ubuntu Server 21.04) and use it with Seafile Server Pro through the terminal only. Whenever I mount the SeaDrive I don’t know what really happens behind the scenes, but SeaDrive is starting very I/O-intensive process, which consumes 95% of my 20GB memory.
My goal really is to list the folders of the Seaseafile (which takes time to see them, especially the largest ones) in order for me to be able to send large amounts of data from other machine using SSH.
Maybe I should be using Fuse for sending data over SSH? I thought fuse is “real-only” and therefore didn’t even try it.
Or maybe there is an idea how to make the SeaDrive eating less memory?
It’s now 13+ hours and the memory is still occupied.
How many files and folders are involved and what kind of disk? I also don’t understand where the 95% memory thing comes from. The screenshot is on io, not memory (?).
From what I see I suspect hard drives and many files.
I’m using Proxmox VE and this is a screenshot from what it looks like now in terms of resources.
It’s about 3TB of data on HDDs (WD Ultrastar DC HC520 He12 12TB), lots of small files.
This doesn’t show anything. Maybe the VM is using 18 GB RAM as cache.
I’m pretty sure the hard drives are your problem. Count with 100-200 iops (per disk, raid will further reduce it) and you’re running SeaDrive and Seafile Server on the same disks. So just one file will probably include fetching an fs object, a data object, storing the two and depending on what happens even more.
Assuming two disks in raid 1 we talk about 75/4 files (<20) per second.
So, do you think that separating seafile and SeaDrive on separate disks would solve the issue potentially? I’m not sure about how proxmox handles the ram. No other process is consuming so much memory.
It would make it faster I think. But without SSDs I don’t think it can be fast.
ok, will reinstall it to ssd! Thanks!