[GH-ISSUE #1867] Plex and Emby Streams and Downloads are Throttled #8797

New Issue

GiteaMirror · 2026-04-30T04:50:50-05:00

GiteaMirror commented

2026-04-30 04:50:50 -05:00

Originally created by @Doubledoros on GitHub (Nov 15, 2025).
Original GitHub issue: https://github.com/fosrl/pangolin/issues/1867

Describe the Bug

I am currently having the same issue for both plex and emby on the most recent version of all programs in the software stack.

I am currently experiencing bandwidth issues with both plex and emby(only two programs that I utilize Pangolin for).
Any attempts to stream anything at speeds higher than 50kb results in constant buffering. As a method of troubleshooting I attempted to download the raw videos from both plex and emby, and still am capped at the same slow speed.

Pangolin stack is currently running on a Racknerd VPS with 1 CPU Core and 2GB RAM. I have done extensive troubleshooting with their tech support, and even had my VPS moved to a different location as a possible measure, but to no avail. I have also confirmed that the DNS records are all correct, and that my host machine is working properly. I have also tried reinstalling the Pangolin stack as a final resort, but that didn't work as well. It appears to be getting software throttled at some point in the tunneling process. Logging has also been disabled in the Pangolin setting menu to see if that helps, but to no avail.

Environment

OS Type & Version: Debian 12
Pangolin Version: 1.12.2
Gerbil Version: 1.2.2
Traefik Version: 3.5
Newt Version: 1.60

To Reproduce

Trying to stream or download any media whether it be video or audio via plex and emby.

Expected Behavior

In the past I was able to stream 4k videos via the Pangolin stack without an issue.

Originally created by @Doubledoros on GitHub (Nov 15, 2025). Original GitHub issue: https://github.com/fosrl/pangolin/issues/1867 ### Describe the Bug I am currently having the same issue for both plex and emby on the most recent version of all programs in the software stack. I am currently experiencing bandwidth issues with both plex and emby(only two programs that I utilize Pangolin for). Any attempts to stream anything at speeds higher than 50kb results in constant buffering. As a method of troubleshooting I attempted to download the raw videos from both plex and emby, and still am capped at the same slow speed. Pangolin stack is currently running on a Racknerd VPS with 1 CPU Core and 2GB RAM. I have done extensive troubleshooting with their tech support, and even had my VPS moved to a different location as a possible measure, but to no avail. I have also confirmed that the DNS records are all correct, and that my host machine is working properly. I have also tried reinstalling the Pangolin stack as a final resort, but that didn't work as well. It appears to be getting software throttled at some point in the tunneling process. Logging has also been disabled in the Pangolin setting menu to see if that helps, but to no avail. ### Environment - OS Type & Version: Debian 12 - Pangolin Version: 1.12.2 - Gerbil Version: 1.2.2 - Traefik Version: 3.5 - Newt Version: 1.60 ### To Reproduce Trying to stream or download any media whether it be video or audio via plex and emby. ### Expected Behavior In the past I was able to stream 4k videos via the Pangolin stack without an issue.

GiteaMirror added the stale label 2026-04-30 04:50:50 -05:00

GiteaMirror closed this issue

2026-04-30 04:50:51 -05:00

GiteaMirror commented

2026-04-30 04:50:52 -05:00

@th3cube commented on GitHub (Nov 21, 2025):

Check this: https://github.com/fosrl/pangolin/issues/1756

I have Plex running with and without newt tunnel and no issues.

So you do the SSL Termination on the VPS and tunnel back to the sam location over Newt ? In that case you would need Split horizon DNS

@th3cube commented on GitHub (Nov 21, 2025): Check this: https://github.com/fosrl/pangolin/issues/1756 I have Plex running with and without newt tunnel and no issues. So you do the SSL Termination on the VPS and tunnel back to the sam location over Newt ? In that case you would need Split horizon DNS

GiteaMirror commented

2026-04-30 04:50:53 -05:00

@Doubledoros commented on GitHub (Nov 22, 2025):

I appreciate the recommendation, but unfortunately that did not work. This was working perfectly while running Pangolin on 1.1x. It seems some aspect of the update to 1.2 is causing the issue.

@Doubledoros commented on GitHub (Nov 22, 2025): I appreciate the recommendation, but unfortunately that did not work. This was working perfectly while running Pangolin on 1.1x. It seems some aspect of the update to 1.2 is causing the issue.

GiteaMirror commented

2026-04-30 04:50:53 -05:00

@jhssilva commented on GitHub (Nov 30, 2025):

I’m currently experiencing the same behavior and have also seen similar reports from other users on Reddit. The problem appears to be related to bandwidth being capped for exposed services, specifically during file downloads or media streaming.

After several hours of debugging, I was able to rule out:
• The WireGuard tunnel (direct tests consistently achieve 70–80 Mbps between homelab ↔ VPS).
• The VPS network itself (speedtests show ~480 Mbps down / ~120 Mbps up).
• Jellyfin or the backend service (direct HTTP requests from inside the Traefik container reach 8–10 MB/s).

The throttling only occurs when accessing the service through Pangolin’s exposed resource.
I also tested exposing the service directly through Traefik (bypassing the resource-based proxy), and the cap still occurs.

@jhssilva commented on GitHub (Nov 30, 2025): I’m currently experiencing the same behavior and have also seen similar reports from other users on Reddit. The problem appears to be related to bandwidth being capped for exposed services, specifically during file downloads or media streaming. After several hours of debugging, I was able to rule out: • The WireGuard tunnel (direct tests consistently achieve 70–80 Mbps between homelab ↔ VPS). • The VPS network itself (speedtests show ~480 Mbps down / ~120 Mbps up). • Jellyfin or the backend service (direct HTTP requests from inside the Traefik container reach 8–10 MB/s). The throttling only occurs when accessing the service through Pangolin’s exposed resource. I also tested exposing the service directly through Traefik (bypassing the resource-based proxy), and the cap still occurs.

GiteaMirror commented

2026-04-30 04:50:54 -05:00

@accik commented on GitHub (Dec 10, 2025):

I seem to have similar issue. I have a VPS (already upgraded to a 4vCPU, 6GB RAM so should be good) that runs Pangolin. My Jellyfin is in my local network (1G up/down) and streams fine with my own Wireguard setup. Only Pangolin seems to hang. No cache between any links.
Checking the resource usage on all steps, only the VPS has some CPU spike but only to max around 20%.
No logs on any part that explain these slowdowns so my suspect is software on Pangolin/Traefik. When testing connection with LibreSpeed, I get 250/100 to the same server that has Jellyfin.

@accik commented on GitHub (Dec 10, 2025): I seem to have similar issue. I have a VPS (already upgraded to a 4vCPU, 6GB RAM so should be good) that runs Pangolin. My Jellyfin is in my local network (1G up/down) and streams fine with my own Wireguard setup. Only Pangolin seems to hang. No cache between any links. Checking the resource usage on all steps, only the VPS has some CPU spike but only to max around 20%. No logs on any part that explain these slowdowns so my suspect is software on Pangolin/Traefik. When testing connection with LibreSpeed, I get 250/100 to the same server that has Jellyfin.

GiteaMirror commented

2026-04-30 04:50:55 -05:00

@jhssilva commented on GitHub (Dec 10, 2025):

I seem to have similar issue. I have a VPS (already upgraded to a 4vCPU, 6GB RAM so should be good) that runs Pangolin. My Jellyfin is in my local network (1G up/down) and streams fine with my own Wireguard setup. Only Pangolin seems to hang. No cache between any links. Checking the resource usage on all steps, only the VPS has some CPU spike but only to max around 20%. No logs on any part that explain these slowdowns so my suspect is software on Pangolin/Traefik. When testing connection with LibreSpeed, I get 250/100 to the same server that has Jellyfin.

Tried on the same server with Caddy and everything works correctly.

@jhssilva commented on GitHub (Dec 10, 2025): > I seem to have similar issue. I have a VPS (already upgraded to a 4vCPU, 6GB RAM so should be good) that runs Pangolin. My Jellyfin is in my local network (1G up/down) and streams fine with my own Wireguard setup. Only Pangolin seems to hang. No cache between any links. Checking the resource usage on all steps, only the VPS has some CPU spike but only to max around 20%. No logs on any part that explain these slowdowns so my suspect is software on Pangolin/Traefik. When testing connection with LibreSpeed, I get 250/100 to the same server that has Jellyfin. Tried on the same server with Caddy and everything works correctly.

GiteaMirror commented

2026-04-30 04:50:56 -05:00

@NeurekaSoftware commented on GitHub (Dec 16, 2025):

Is this a genuine issue affecting media servers?

For those with bandwidth issues with Plex: That’s only when accessing via a reverse proxy through Pangolin (Traefik) right?

I am just curious because I use Pangolin. I haven’t noticed issues but if there are bandwidth problems I definitely will consider moving back to Caddy.

@NeurekaSoftware commented on GitHub (Dec 16, 2025): Is this a genuine issue affecting media servers? For those with bandwidth issues with Plex: That’s only when accessing via a reverse proxy through Pangolin (Traefik) right? I am just curious because I use Pangolin. I haven’t noticed issues but if there are bandwidth problems I definitely will consider moving back to Caddy.

GiteaMirror commented

2026-04-30 04:50:57 -05:00

@accik commented on GitHub (Dec 17, 2025):

Tried on the same server with Caddy and everything works correctly.

Thanks for the reply. Were you able to switch from Traefik to Caddy on Pangolin or did you go through a separate solution? Any guides or references are appreciated, but I would like to stay on Pangolin.
I looked at Caddy and seems quite easy solution, might test it out later.

@accik commented on GitHub (Dec 17, 2025): > Tried on the same server with Caddy and everything works correctly. Thanks for the reply. Were you able to switch from Traefik to Caddy on Pangolin or did you go through a separate solution? Any guides or references are appreciated, but I would like to stay on Pangolin. I looked at Caddy and seems quite easy solution, might test it out later.

GiteaMirror commented

2026-04-30 04:50:57 -05:00

@jhssilva commented on GitHub (Dec 18, 2025):

Tried on the same server with Caddy and everything works correctly.

Thanks for the reply. Were you able to switch from Traefik to Caddy on Pangolin or did you go through a separate solution? Any guides or references are appreciated, but I would like to stay on Pangolin.
I looked at Caddy and seems quite easy solution, might test it out later.

Used caddy as main one. Leveraged some predefined routes for streaming services, and then passed the traffic to pangolin for the rest of the services.
For guides, unfortunately I don't have something at hand. I've read the docs from the caddy, and how to leverage the rest of the ports, and spent 1 day on this.

@jhssilva commented on GitHub (Dec 18, 2025): > > Tried on the same server with Caddy and everything works correctly. > > Thanks for the reply. Were you able to switch from Traefik to Caddy on Pangolin or did you go through a separate solution? Any guides or references are appreciated, but I would like to stay on Pangolin. > I looked at Caddy and seems quite easy solution, might test it out later. Used caddy as main one. Leveraged some predefined routes for streaming services, and then passed the traffic to pangolin for the rest of the services. For guides, unfortunately I don't have something at hand. I've read the docs from the caddy, and how to leverage the rest of the ports, and spent 1 day on this.

GiteaMirror commented

2026-04-30 04:50:58 -05:00

@AstralDestiny commented on GitHub (Dec 25, 2025):

Make sure you don't hairpin any of the hops it'll kill your performance and remember your upload where newt is, is the main factor. As it has to upload up to the vps if you are using vps and the vps can only download as fast as you can upload to it.

speedtest.net tests or similar are not reliable to actually verify connections as you are connecting to a service that is just bursting, over doing sustained connections and with overhead from encryption and a tunnel passthrough through hops.. If you want to test, Pass a iperf3 to the host down the tunnel to the newt endpoint. and then leave it running, If while running and you get a dip it can either be an over burderned vps or if you don't get a reliable speed then it's the networking as it leaves your vps.

@AstralDestiny commented on GitHub (Dec 25, 2025): Make sure you don't hairpin any of the hops it'll kill your performance and remember your upload where newt is, is the main factor. As it has to upload up to the vps if you are using vps and the vps can only download as fast as you can upload to it. speedtest.net tests or similar are not reliable to actually verify connections as you are connecting to a service that is just bursting, over doing sustained connections and with overhead from encryption and a tunnel passthrough through hops.. If you want to test, Pass a iperf3 to the host down the tunnel to the newt endpoint. and then leave it running, If while running and you get a dip it can either be an over burderned vps or if you don't get a reliable speed then it's the networking as it leaves your vps.

GiteaMirror commented

2026-04-30 04:50:59 -05:00

@github-actions[bot] commented on GitHub (Jan 9, 2026):

This issue has been automatically marked as stale due to 14 days of inactivity. It will be closed in 14 days if no further activity occurs.

@github-actions[bot] commented on GitHub (Jan 9, 2026): This issue has been automatically marked as stale due to 14 days of inactivity. It will be closed in 14 days if no further activity occurs.

GiteaMirror commented

2026-04-30 04:50:59 -05:00

@frigi83 commented on GitHub (Jan 9, 2026):

Hey there! I just got Pangolin set up myself, and I'm super excited to get Plex exposed with it! I've been using Traefik for a few years now, so I'm pretty familiar with it.

I've got a pretty basic VPS (just 1 vCPU, 2GB RAM), and I noticed things getting a bit choppy when I started streaming. Thanks to btop, I pinpointed the bottleneck: Crowdsec was hogging 100% of my CPU (normally it idles around 1-3%). As soon as I disabled the Crowdsec middleware, everything smoothed out beautifully, and my CPU usage rarely goes above 20%! 🎉

@frigi83 commented on GitHub (Jan 9, 2026): Hey there! I just got Pangolin set up myself, and I'm super excited to get Plex exposed with it! I've been using Traefik for a few years now, so I'm pretty familiar with it. I've got a pretty basic VPS (just 1 vCPU, 2GB RAM), and I noticed things getting a bit choppy when I started streaming. Thanks to `btop`, I pinpointed the bottleneck: Crowdsec was hogging 100% of my CPU (normally it idles around 1-3%). As soon as I disabled the Crowdsec middleware, everything smoothed out beautifully, and my CPU usage rarely goes above 20%! 🎉

GiteaMirror commented

2026-04-30 04:51:00 -05:00

@github-actions[bot] commented on GitHub (Jan 25, 2026):

This issue has been automatically marked as stale due to 14 days of inactivity. It will be closed in 14 days if no further activity occurs.

@github-actions[bot] commented on GitHub (Jan 25, 2026): This issue has been automatically marked as stale due to 14 days of inactivity. It will be closed in 14 days if no further activity occurs.

GiteaMirror commented

2026-04-30 04:51:01 -05:00

@github-actions[bot] commented on GitHub (Feb 8, 2026):

This issue has been automatically closed due to inactivity. If you believe this is still relevant, please open a new issue with up-to-date information.

@github-actions[bot] commented on GitHub (Feb 8, 2026): This issue has been automatically closed due to inactivity. If you believe this is still relevant, please open a new issue with up-to-date information.

Sign in to join this conversation.

Branches Tags

main

dev

dependabot/npm_and_yarn/npm-dependencies-8b2d4a9f3a

local-connection

dependabot/go_modules/install/go-install-dependencies-3804ca7238

dependabot/docker/docker-dependencies-4faa477378

dependabot/github_actions/github-actions-dependencies-6d79802a48

api-improvements

feat/remember-last-idp-on-smart-login-form

refactor/batch-status-requests

dependabot/npm_and_yarn/multi-5f1280885e

fix/non-semver-version-error

private-resource-page

resource-launcher

backhaul

exit-node-reconnect

feat/command-palette

ssh

delete-account

msg-delivery

org-only-idp

cicd

patch

site-targets-auto-login

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: github-starred/pangolin#8797