Performance drop after migration: High %WA and Load Average at 550+ connections (Dropbox VFS)

Help and Support
1

Hello everyone!

I’m looking for some help regarding my Rclone setup. I use Rclone to link my server with Dropbox, using a dedicated API client for each mount.

With my current configuration, I used to handle up to 1,200 concurrent connections on my previous server. However, after migrating to a new machine, I can't seem to get past 500–550 connections.

Server Specs:

  • CPU: 2x Intel Xeon E5-2697A (40 cores / 80 threads)

  • RAM: 256GB

  • Network: 10Gbps

  • Storage: 2TB NVMe for VFS Cache.

Current Mount Flags: --allow-other --allow-non-empty --vfs-cache-mode full --buffer-size 32M --vfs-cache-max-age 9999h --tpslimit 12 --transfers 32 --checkers 32 --read-only --no-modtime --no-checksum --no-seek --no-traverse

Cache Strategy: I am using vfs-cache-mode full on the 2TB NVMe. For the mount with the highest traffic (Movies), I have assigned a 120GB RAM disk for the cache directory.

The Problem: As soon as the server hits approximately 550 concurrent connections, the I/O Wait (%WA) spikes, the Load Average skyrockets, and the server eventually collapses/freezes.

On my old server, this didn't happen with the same flags. Could you give me some advice on what might be causing this bottleneck or if there's something wrong with my current flags?

Thanks in advance for your help!

1 Like
2

I would test the speed outside of rclone, on both servers.

A good tool is Cloudflare's dashboard here

There's also this third party CLI

3

welcome to the forum,

hard to help as you did not answer the questions in the help and support template??

does nothing on a mount, so remove that.

really need to be 1000% sure that is absolutely required.
as it allow for over mounting and unintended consequences.

4

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.