New SNO, need help :)

Hello everyone, I’m just starting my sno journey. Running 2 nodes via docker I see a constant traffic of 250mbps. This is normal? The information panel also works inexplicably. Is it possible to transfer metrics to Graphana? Is it possible to store several nodes within one IP address? Because I have a disk shelf with 68 HDDs. Therefore, there are no problems with growth. Will the load on the return channel increase as the nodes are filled? What are the limitations at all. thank you

1 Like

This is currently above-normal: as performance testing is underway. But Storj is chasing at least one new large customer: so we’re all hoping it becomes normal :money_mouth_face:

There are a couple common Graphana configs: check here for one. And other ways to see everything in one place.

Yes, you can have many nodes behind one IP (like on different ports). However Storj has all nodes in the same /24 subnet share uploads. That will at least mean nodes behind your single IP all share… but also may mean you’re sharing ingress with your neighbours (or others using the same VPN service+endpoint). So if you do have access to more different /24 IPs you can grow your used-diskspace faster. (check here to see how many other nodes you’re sharing with now)

If I had to estimate… and ignore the recent burst of uploads for testing… I’d say a 10TB HDD would normally take 2-2.5 years to fill. But again currently things are moving much faster: and other SNOs will have their own estimates.

68 HDDs will work fine - my guess is you’d probably need 64GB of RAM if you’re using Linux+Docker. (some SNOs have 100-150+ HDDs per server). The most straightforward config would be to give every node one disk and a different port. However it will take a long time to fill them if you don’t have a way to use more IP addresses. Most smaller setups will have people fill one-disk-at-a-time… then add space when needed.

Some SNOs virtualize, some run various RAID levels, some use fancy SSD caching layers or ZFS metadata devices. But for Linux the vanilla option is one-disk-per-node in Docker, EXT4 with noatime. I like to keep it simple.

Welcome, and have fun!

1 Like

thank you, my friend, will my nodes, which are full, become empty in the near future and I will have to start all over again? That’s what I’m currently doing, docker and node in ext4.

I have this error Z INFO failed to sufficiently increase send buffer size (was: 208 kiB, wanted: 2048 kiB, got: 416 kiB). See UDP Buffer Sizes · quic-go/quic-go Wiki · GitHub for details. {“Process”: “storagenode”}

But buffer 7.5 mb, i use linux mint

You’ll want to add the following to /etc/sysctl.conf

net.ipv4.tcp_fastopen=3
net.core.rmem_max=2500000
net.core.wmem_max=2500000
2 Likes