New node - getting “connection reset by peer” on all satellites - tried everything I can think of
Hey everyone, I’ve been banging my head against this for 2 days straight and I’m completely stuck. Would really appreciate some help here.
My setup
Running a fresh node on a MeLE Cyber X1 mini PC (Intel N150) with Windows 11. Installed using the official MSI installer, version v1.150.3. Got a 14TB WD My Book on USB for storage (NTFS, 11.5TB allocated). ISP is A1 Bulgaria with a Technicolor TC7200 modem/router.
External address is set to alexstorj.duckdns.org:28967
What’s happening
Every single satellite ping fails with “connection reset by peer”. It’s been like this since I set it up — the node has never successfully contacted any satellite:
ERROR contact:service ping satellite failed
"error": "ping satellite: failed to ping storage node, your node indicated error code: 0,
rpc: tcp connector failed: rpc: read tcp 10.54.6.4:42996->213.191.163.143:28967:
read: connection reset by peer"
All 4 satellites (eu1, us1, ap1, saltlake) give me the same error.
What actually works
The weird thing is — the node is clearly reachable from the outside. When I check port 28967 on yougetsignal.com it shows open. When I access the IP from a different PC trough HTTP in a browser, I get a response back (empty page or the DRPC JSON message). The dashboard loads fine on localhost:14002 and shows the node is running but offline with QUIC Misconfigured.
The node starts up fine, creates all it’s databases, and reaches out to satellites. It’s only when the satellites try too connect back that it fails.
All the network configs I’ve tried
I went pretty deep on this one. My setup has an A1 Technicolor modem/router and behind it a TP-Link TL-WR940N.
With TP-Link in router mode (it was getting its own public IP via portbase passthrough):
-
Set up port forwarding for 28967 TCP+UDP on the TP-Link
-
Connection reset by peer
With TP-Link in Access Point mode and MeLE getting a direct public IP (NO NAT at all):
-
The passthrough feature on the Technicolor gives the MeLE a public IP directly
-
No port forwarding needed at all — completely open to the internet
-
Still connection reset by peer
So it’s not a NAT issue and its not a port forwarding issue.
Software stuff I’ve tried
-
Turned off Windows Firewall completely — same result
-
Added Windows Defender exclusions for the Storj folders and the storagenode.exe process
-
Disabled SPI Firewall on the TP-Link
-
Disabled DoS protection on the TP-Link
-
Tried Docker first (had USB drive mount issues with WSL2), then native binary, then the MSI installer — all gave me the same connection reset
-
Tested with v1.151.1-rc and v1.150.3 — no difference
-
No Hyper-V adapters leftover from Docker (checked with ipconfig /all)
-
No VPN, no extra antivirus besides Defender
Identity
identity.cert: 2 x BEGIN markers
ca.cert: 1 x BEGIN marker
I know the auth token step isnt required anymore per the docs.
Netstat while node is running
TCP 0.0.0.0:28967 0.0.0.0:0 LISTENING
TCP [::]:28967 [::]:0 LISTENING
UDP 0.0.0.0:28967 *:*
UDP [::]:28967 *:*
The thing that confuses me most
Regular HTTP connections from outside reach the node no problem. It’s specifically the DRPC/TLS handshake from the satellites that gets reset. To me this looks like something is messing with TLS connections specifically but not blocking the port.
Could this be the ISP hardware (Technicolor TC7200) doing some kind of deep packet inspection even in pass-through mode? Or is there something else on Windows that could interfere with TLS handshakes that I’m missing?
I’ve rebooted everything multiple times, reinstalled the node from scratch 3 times, tested with and without every firewall and NAT config I can think of. I’m out off ideas.
Any help would be hugely appreciated. Thanks!
