Issues with QUIC Configuration on Multiple Storj Nodes on the Same Machine

Node Operators troubleshooting
1

I’ve set up three Storj nodes on the same machine and assigned three different public IP addresses to each node. Here is my configuration:

Node 1:

docker run -d --restart unless-stopped --stop-timeout 300 \
    -p 28901:28967/tcp \
    -p 28901:28967/udp \
    -p 127.0.0.1:14001:14002 \
    -e WALLET="0xXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX" \
    -e EMAIL="xxxxxxxxxx@gmail.com" \
    -e ADDRESS="xx.xx.xx.21:28901" \
    -e STORAGE="13.1TB" \
    --user $(id -u):$(id -g) \
    --mount type=bind,source="/storj/identity/storagenode",destination=/app/identity \
    --mount type=bind,source="/storj/data",destination=/app/config \
    --name storagenode storjlabs/storagenode:latest

Node 2:

docker run -d --restart unless-stopped --stop-timeout 300 \
    -p 28902:28967/tcp \
    -p 28902:28967/udp \
    -p 127.0.0.1:14002:14002 \
    -e WALLET="0xXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX" \
    -e EMAIL="xxxxxxxxxx@gmail.com" \
    -e ADDRESS="xx.xx.xx.22:28902" \
    -e STORAGE="13.1TB" \
    --user $(id -u):$(id -g) \
    --mount type=bind,source="/storj2/identity/storagenode2",destination=/app/identity \
    --mount type=bind,source="/storj2/data",destination=/app/config \
    --name storagenode2 storjlabs/storagenode:latest

Node 3:

docker run -d --restart unless-stopped --stop-timeout 300 \
    -p 28903:28967/tcp \
    -p 28903:28967/udp \
    -p 127.0.0.1:14003:14002 \
    -e WALLET="0xXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX" \
    -e EMAIL="xxxxxxxxxx@gmail.com" \
    -e ADDRESS="xx.xx.xx.23:28903" \
    -e STORAGE="15TB" \
    --user $(id -u):$(id -g) \
    --mount type=bind,source="/storj3/identity/storagenode3",destination=/app/identity \
    --mount type=bind,source="/storj3/data",destination=/app/config \
    --name storagenode3 storjlabs/storagenode:latest

Despite this setup, I am receiving the following error for Nodes 2 and 3:

“QUIC is misconfigured. You must forward port 28967 for both TCP and UDP to enable QUIC. See Step 3. Setup Port Forwarding - Storj Docs on how to do this.”

Node 1 seems to be working fine without this error.
Nodes 2 and 3 are using different ports for communication (28902 and 28903).

I have ensured that port forwarding is set up correctly on my router for the respective ports and IP addresses. Each node is running on a different public IP, so I’m not sure why this misconfiguration message is appearing.

Questions:

  1. Is there a specific requirement for QUIC configuration that I might be missing for nodes running on different ports?
  2. How can I ensure that QUIC is configured correctly for multiple nodes on the same machine with different IP addresses?

Any insights or suggestions would be greatly appreciated!

2

How long have you waited? I often see a node be Online after startup… but QUIC says Misconfigured… then I check back in an hour and they’re both OK. (I think the Satellites’ online-check and quic-check may be at different times?)

Edit: It looks like this page lets you manually run a QUIC check?

1 Like
3

About 10+ months ago, Go bumped their lib for UDP use, screwing up STORJ’s implementation - breaking some backward compatibility.

1.) You can completely ignore this problem as UDP still isn’t really used.
2.) I would guess you can correct the Quick issue by specifying a specific unique adapter/pathway IP per instance. ie: -p 192.168.1.10:28901:28967, but I don’t run Docker- YMMV.
3.) If you change start order of your nodes, it should notice: it’s the first one started & subsequently confirmed that gets the quick OK.

GL

1 cent
2.) * Credit @arrogantrabbit, maybe it’s enough to change the internal port & your start params to match ie: 192.168.1.11:28901:28901 … Betting AR can confirm.

1 Like
4

Did you configure the buffer size as in Linux Configuration for UDP - Storj Docs ? Check your storagenode log on startup, there is a warning if buffer size is too small. I use net.core.rmem_max=7500000 with multiple docker nodes all OK, same docker parameters as you.

2 Likes
5

any chance it’s a firewall port forward problem? ports 28901 and 28903 may not be making it from your router to your node box?

1 Like