Hi,

So just as a brief setup situation:

10Gb Internet line

using a 10Gb router using fiber connected to a 10Gb aggregation switch where the server is connected.

SO fresh install of Unraid using 1x 10Gb link on the server (no bonding).

If I do an iperf3 between this Unraid server to another server within the same network it does hit an avg of 8Gbit/s both In & Out, so that leaves out an issue with the LAN.

but the moment I do either a direct speedtest through ookla or speed tracker apps in unraid. All tests comes out with a result under 1Gbit/s (avg 700-800mbps) for both Upload and Download.

This is all consistent when doing the internet speedtest through Speedtest Tracker app, Ookla CLI through the Terminal, a Windows VM (despite the nic being 10Gbe).

When doing the internet speedtest outside of Unraid, in this case my normal windows PC, I would hit around 8.5Gbps as my PC is also connected with a 10Gb link.

any idea what would the issue be or better way on testing the internet speed through unraid?

as far as firewall goes, it has the same policy as any other 10gb systems I have within my network and all of them are able to reach speeds above 1Gb

please post a diag...

as example for 2.5 gb there is an additional things you need to do and run to make sure your getting and using the full link speed:
example:
https://forums.unraid.net/topic/141349-plugin-realtek-r8125-r8126-r8168-and-r815267-drivers/#findComment-1279775

there are some known issues and some advance configuration you may need to do
https://www.youtube.com/watch?v=ZMMopEiwiEo
https://www.youtube.com/watch?v=25umVbN__6M

the diag would have helpd and been quicker... so your on you own after this line of testing....
1st check your ip route to make sure your standard 1 GB eth nic and your 10GB connection is set to use teh 10 GB link as this is a common gothca

run ip route and look at your subnet lan to the dev interface adn MAKE SURE IT IS TEH INTERFACE THAT IS 10 GB link...

ok I have to asume eth0 sine no diag, no hardware. coper?sfp?sfp+ etc...

so you will need to fix the comands and interfaces...

A. ) Check Link Status & Speed

# Show all interfaces and their link speeds

ethtool eth0

# If multiple NICs exist, adjust eth0 → eth1/eth2

Look for:

• Speed: 10000Mb/s

• Duplex: Full

• Link detected: yes

If it shows 1000Mb/s, you’re capped at 1 GbE somewhere (driver, switch port, SFP, or cable).

B. ) Verify IP and Interface

ip addr show eth0

ip -s link show eth0

This confirms the IP assignment, MTU, and packet errors/drops.

Optional but I'd set it.... as your mtu may still be 1500 for 1GB
C.) Test Jumbo Frames (optional if your network supports >1500 MTU)

# Check current MTU

ip link show eth0

# Temporarily set to 9000 (for test)

ip link set eth0 mtu 9000

# Ping with large packets

ping -M do -s 8972 <gateway_or_other_host>

D.) Check Driver & Module

ethtool -i eth0

lspci -nnk | grep -A3 -i eth

This tells you which driver is in use. Some 10 GbE NICs (Mellanox, Intel X550, Broadcom) need firmware or tuning.

(Dirver plugin for melnox exist as a plugin for example you may need adatioanl drivers/modprobe as teh unraid kerneal may not have a driver for your hardware...)

E.) Monitor Traffic or somethign that can I'm used to deb and having deb coanads or something I can use to monitor...
iftop -i eth0

nload eth0

unraid cand do this with adatial 3rd party slack pack install..

look into: (to install: slackpkg install iftop or via NerdTools plugin in Unraid).

F.) Systemctl / Service Checks

Unraid doesn’t use NetworkManager by default but you can still check lets review some still worth a mention... unraid usrs systemvar and scripts located elswhere...

# See if networking service is running

systemctl status networking

# Restart interface

ifdown eth0 && ifup eth0

Since iperf3 works well inside LAN, try:

# Run a multithreaded speedtest (CLI)

speedtest --servers

speedtest --threads=8 --progress=no

If you don’t have Ookla CLI

curl -s https://install.speedtest.net/app/cli/install.rpm.sh | bash

and lastly if you've wathed the youtbe and still need to use tcp staks and settings..
TCP Stack & Offload Settings

# Check NIC offloading

ethtool -k eth0

# Disable problematic offloads (test only)

ethtool -K eth0 tso off gso off gro off

Sometimes GRO/TSO offload issues cause capped speed in virtualized or container workloads.

so again please post a diag so we can get a better picture of what hardware and linux sytem data..
Recap:

1. Run ethtool eth0 and confirm the link speed.

2. Run ethtool -k eth0 to see offload features.

3. Try a multithreaded Ookla test (--threads=8).

On 9/20/2025 at 5:53 PM, bmartino1 said:

please post a diag...

as example for 2.5 gb there is an additional things you need to do and run to make sure your getting and using the full link speed:
example:
https://forums.unraid.net/topic/141349-plugin-realtek-r8125-r8126-r8168-and-r815267-drivers/#findComment-1279775

there are some known issues and some advance configuration you may need to do
https://www.youtube.com/watch?v=ZMMopEiwiEo
https://www.youtube.com/watch?v=25umVbN__6M

the diag would have helpd and been quicker... so your on you own after this line of testing....
1st check your ip route to make sure your standard 1 GB eth nic and your 10GB connection is set to use teh 10 GB link as this is a common gothca

run ip route and look at your subnet lan to the dev interface adn MAKE SURE IT IS TEH INTERFACE THAT IS 10 GB link...

ok I have to asume eth0 sine no diag, no hardware. coper?sfp?sfp+ etc...

so you will need to fix the comands and interfaces...

A. ) Check Link Status & Speed

# Show all interfaces and their link speeds

ethtool eth0

# If multiple NICs exist, adjust eth0 → eth1/eth2

Look for:

• Speed: 10000Mb/s

• Duplex: Full

• Link detected: yes

If it shows 1000Mb/s, you’re capped at 1 GbE somewhere (driver, switch port, SFP, or cable).

B. ) Verify IP and Interface

ip addr show eth0

ip -s link show eth0

This confirms the IP assignment, MTU, and packet errors/drops.

Optional but I'd set it.... as your mtu may still be 1500 for 1GB
C.) Test Jumbo Frames (optional if your network supports >1500 MTU)

# Check current MTU

ip link show eth0

# Temporarily set to 9000 (for test)

ip link set eth0 mtu 9000

# Ping with large packets

ping -M do -s 8972 <gateway_or_other_host>

D.) Check Driver & Module

ethtool -i eth0

lspci -nnk | grep -A3 -i eth

This tells you which driver is in use. Some 10 GbE NICs (Mellanox, Intel X550, Broadcom) need firmware or tuning.

(Dirver plugin for melnox exist as a plugin for example you may need adatioanl drivers/modprobe as teh unraid kerneal may not have a driver for your hardware...)

E.) Monitor Traffic or somethign that can I'm used to deb and having deb coanads or something I can use to monitor...
iftop -i eth0

nload eth0

unraid cand do this with adatial 3rd party slack pack install..

look into: (to install: slackpkg install iftop or via NerdTools plugin in Unraid).

F.) Systemctl / Service Checks

Unraid doesn’t use NetworkManager by default but you can still check lets review some still worth a mention... unraid usrs systemvar and scripts located elswhere...

# See if networking service is running

systemctl status networking

# Restart interface

ifdown eth0 && ifup eth0

Since iperf3 works well inside LAN, try:

# Run a multithreaded speedtest (CLI)

speedtest --servers

speedtest --threads=8 --progress=no

If you don’t have Ookla CLI

curl -s https://install.speedtest.net/app/cli/install.rpm.sh | bash

and lastly if you've wathed the youtbe and still need to use tcp staks and settings..
TCP Stack & Offload Settings

# Check NIC offloading

ethtool -k eth0

# Disable problematic offloads (test only)

ethtool -K eth0 tso off gso off gro off

Sometimes GRO/TSO offload issues cause capped speed in virtualized or container workloads.

so again please post a diag so we can get a better picture of what hardware and linux sytem data..
Recap:

1. Run ethtool eth0 and confirm the link speed.

2. Run ethtool -k eth0 to see offload features.

3. Try a multithreaded Ookla test (--threads=8).

thanks for the detailed post, however what I did first was just deleted my network config in the unraid config folder. for it to create a new one and somehow that fixed my issue :)

Is this like chatGPT or something? slackpkg isn't a command that runs on unraid and NerdTools is deprecated in unraid 7

On 9/20/2025 at 5:53 PM, bmartino1 said:

E.) Monitor Traffic or somethign that can I'm used to deb and having deb coanads or something I can use to monitor...
iftop -i eth0

nload eth0

unraid cand do this with adatial 3rd party slack pack install..

look into: (to install: slackpkg install iftop or via NerdTools plugin in Unraid).

slackpkg does exist and can run... But, I would refer to using unget to install per work with nerdtools at the end... unraids slackpkg doen't have a repo set... unraid makes atempts to disable it for stabiltiy... i gave taht as an example not as a flol along ...

go find you rbinary and thro it in the extra folder...

...no ai tools like grammly adn lanuage tool to help corect other issues...

Edited October 7, 2025Oct 7 by bmartino1
correction and edits

23 hours ago, bmartino1 said:

slackpkg does exist and can run... But, I would refer to using unget to install per work with nerdtools at the end... unraids slackpkg doen't have a repo set... unraid makes atempts to disable it for stabiltiy... i gave taht as an example not as a flol along ...

go find you rbinary and thro it in the extra folder...

...no ai tools like grammly adn lanuage tool to help corect other issues...

I have 3 unraid servers, slackpkg does not exist on any of them.

root@:~# slackpkg

bash: slackpkg: command not found

root@:~#

You didn't say un-get, you said nerdtools.

Please don't put out bad info then attack me like I'm the problem.

you are the probelm. CLealry you didn't read any of the nerdtools post nor how this coamnd exist and works. also don't take over someone else post and thread make your own general post!

Please be nice to each oher and create a new support topic instead of reviving an old one.