Metric Agent connects but HammerDB GUI immediately aborts the display session (Windows 11, v5.0)

[Tfonseca03]

Hello,
I've only recently started learning how to use this tool, so sorry if I'm missing something obvious and the solution turns out to be a simple fix.
I'm having an issue with the Metric Agent on HammerDB 5.0 (Windows 11).

The Metric Agent starts normally and the GUI successfully connects to it. Metrics begin to stream, but after a few seconds the HammerDB GUI immediately closes the connection.

Here is the sequence reproduced every time:

Any idea what might be causing this?
I'm running it on Windows 11, and I’ve tried launching both hammerdb.exe and agent.exe in administrator mode.
This happens with SQLSERVER

[sm-shaw]

hi, when we've seen reports of things like this before it is because of 3rd party tools such as bitdefender - please check there are no 3rd party tools closing the ports that hammerdb opens as it says closed by request.

[Tfonseca03]

Hi, thanks for your help.

I’ve tested everything you suggested:

Fully uninstalled McAfee (including leftover services and drivers)
Fully uninstalled ProtonVPN (including its TAP/WireGuard drivers)
Restarted the PC multiple times
Confirmed that no 3rd-party firewall, VPN or AV services are running
Verified that Windows Firewall allows HammerDB and agent.exe
Checked that the agent process stays alive until the moment the GUI tries to display metrics

After all of this, I still get the exact same behavior:
I can start the Metric Agent manually or automatically
The agent stays open without closing
As soon as I click Metrics → Display Metrics in the GUI, the agent immediately terminates the connection

Also is there any alternative way to use the metrics without relying on the GUI Display panel
For example:
Using metrics through the CLI only, because I need the metrics to jobs section

Thanks again for your guidance.

[sm-shaw]

So here is an example of what should happen, start the agent, put the id (ie. port) and hostname in the options and connect:

and the metrics data is shown in the display, in this example I am connecting to Windows across the network from a Linux display.

If this is closing when it connects, then it is getting the windows socket error WSAECONNABORTED, if you search for this you can see that firewall or security software is typically the cause.

Once the agent is running try connecting with powershell test-netconnection

Also if you have it use a tool such as telnet to connect (putty also has telnet) and see if a connection can be made, it should stay open rather than close immediately.

$ telnet raven 22384
       Trying 192.168.1.141...
       Connected to raven.home.
       Escape character is '^]'.

Also check the windows logs to see if there are any related messages to the connection being closed.

And yes the metstart connection if running locally will run the metrics in the CLI and start the local agent for you, however this uses exactly the same agent as the GUI, so is likely to encounter the same problems.

We have not been able to reproduce this, so is almost certainly a security setting somewhere that is closing the port.

[Tfonseca03]

Hi, following up with more details on the Metrics Agent disconnect issue.

I have removed all VPN-related software that could interfere with TCP connections (ExpressVPN is fully uninstalled).
The only VPN/security product still installed is Check Point (university vpn). I’m wondering if its drivers or filters could still be causing a conflict, even though it is not actively connected.

I researched the WSAECONNABORTED error extensively and tested everything I could:
Telnet to the Metrics Agent port works perfectly — it opens and stays open without closing
The Metrics Agent itself starts normally and accepts connections
I only get the disconnect inside HammerDB (both GUI and CLI) when I run Display or metstart
The issue happens identically in GUI and CLI
In the cli appears "Received a new display request from host fe80::439f:379d:a0f4:9e34%21" it is normal?

Questions:

Could Check Point (even when not connected) still interfere with HammerDB’s raw TCP display channel?
Is there a way to force HammerDB GUI/CLI to use only IPv4 and ignore IPv6 link-local addresses (fe80::), since that seems to trigger the duplicate display request and abort?
Any guidance would be greatly appreciated.

[sm-shaw]

Yes for IPv4 try using 127.0.0.1 or localhost to connect to an agent started locally.
Yes check point could still be enforcing rules even when not connected to close ports of applications it does not recognise.
Note that there are 2 ports involved here to send and receive as the agent can also run a different system.
Also check the windows logs, it should be noted in a log if a port was closed.