The strange case of bad performance unveiled

For the longest time I’ve had these issues with streaming from my SMB-based shares on my file server but I always had a hard time figuring out when the problems first started occurring.

As a long time infuse user, performance (with this same setup) was always good, and all of a sudden (at least I thought), it became SLOW.

I followed all links to threads I could find, confirming my server with SMBv3, using Jumbo frames, and everything else I could think of. The problem was always there. And it wasn’t a bad network either. I do know a thing or two about setting up a network all my critical devices are wired.

After almost a year I finally was able to find and reproduce the problem, and the goal with this thread is to provide my finding to the community as well as asking the developers to see if this is something that can be addressed.

So, in short, my problem was caused by my file server being listed in Infuse twice, with different configurations. One was the server with its LAN IP; the other was same server but with its Tailscale IP. Tailscale is the VPN setup I use to access my devices when traveling.

When I recently moved to using Tailscale Subnets I didn’t need to have the client installed in my fileserver anymore, as I can access it even when traveling via the IP address of the my LAN. And as soon as I removed the second listing of that server (the one with the Tailscale IP), Infuse immediately started working with its excellent performance that I was so used to back in the day before - now with the benefit of hindsight - I added my VPN IP to it.

It’s as if Infuse somehow gets confused by the double listing. Any thoughts of why this might be?

Personally I won’t suffer from this issue anymore but I feel it’s still relevant to discuss and bring to light.

TL;DR: Infuse gets slow in streaming when the same source is listed twice, even though it is theoretically a different source (same server but with different IP).

Steps to reproduce (there are many but this is my exact setup):

  • Add ServerA as a source to Infuse via IP address
  • Install Tailscale in ServerA
  • Add ServerA to Infuse via the Tailscale IP (which obviously will only work when VPN is on)
  • Run Infuse for a few days
  • After a while, performance on the LAN connection will result in 20+ seconds loading time and overall bad performance

P.S.: It doesn’t matter where playback will take place. The bad performance will happen on wired or wireless. From Apple TV (wired) to desktop computers. All Infuse playback will be affected by this.

1 Like

Cool that you shared. :+1:t3:

1 Like

Thanks for sharing.
I’m getting ready to replace Spectrum with TMHI (T-Mobile Home Internet) and is external facing IPV6. My Channels DVR server (Windows) has Tailscale built in. Plex is on Linux. I’ve registered the Linux server as well as the Channels DVR service.

You’re welcome! I would love if the developers pitched in on this though; Seems like it’s something that could be easily fixed. I won’t say it’s unusable with this “dual homed”-kinda of situation but it’s a really bad experience.

Additionally, this is also just a test I ran and I could be wrong. My Infuse experience is back to what it’s supposed to, but I’ve seen stranger things happening.

1 Like

Bump!

I’d really love for this to be seen or acknowledge by the dev team. Alternatively, for anyone who ends up in this thread because they might be using Tailscale and Infuse: I recently started using the subnet routers feature of Tailscale (where a single Tailscale host in your network can route traffic to other non-Tailscale hosts) and this feature is awesome.

It will also allow you to use Infuse when away without changing a single thing! Just browse your library like you would if you were inside your home network. The Tailscale client (where you’re browsing your network from) will route the traffic to the host in your network that has Tailscale enabled and it will serve as a relay between your Infuse and wherever your hosting your library at.