GlusterFS Replication Question (Distributed / Replicated Storage) for increasing streaming bandwidth

Help and Support
1

Hello all,

I was wondering is someone could help me figure something out. I asked ChatGPT whether it would theoretically be possible to increase streaming bandwidth if I were to use GlusterFS with multiple nodes.

The idea is that I have large video files hosted on OneDrive/SharePoint and I would like to increase absolute bandwidth when streaming one of these files via Plex. While my current setup works flawlessly, I just had the idea yesterday of bundling multiple simultaneous connections to the same video file via GlusterFS.

The idea is that I would mount the same remote four times into four different folders and create a GlusterFS volume with all four folders. Apparently, GlusterFS can then bundle these four requests, increasing bandwidth.

So my question is, does this work and or does this make sense?

Best,
Daniel

2

I think its is question to GlusterFS forum.

3

I'd say give it a try and see how it works out as I've not used that with rclone before.

I wonder if other rclone users have tried it out and think it's a great forum question!

4

What an honour to have one of the rclone kings reply. Thanks so much for your continued contributions here, you helped me out so much without even knowing.

I tried an initial run, apparently GlusterFS doesn't seem to like directories with rclone mounts behind them, even tried using cache-mode full without improvement. I admittedly did not put more than 30minutes into it.

I do think I may be onto something here, SharePoint does cap bandwidth per request, so maybe this is a way to circumvent that with something like GlusterFS.

If anyone has tried this, please do share.

PS: the issue with GlusterFS and rclone mounts seems to be one of extended attributes, unfortunately this exceeds my knowledge of filesystems.

5

I feel like everything that @random404 has ever posted if anyone has tried this, that would the person to ask.

6

I gave up on GlusterFS long time ago, they always have some weird sync issues that you then spend a lot of time debugging.

And will not give what you are looking for in fact could make it even worse.

If you just want maximum read speeds just use the rclone vfs local cache on nvme disks and use something like zfs to pool more disks if needed for more storage