[Question] Questions about blockchain node operation

[victor0807]

Hello, I'm victor

Thank you for helping before. I have a issue about maintain Sei Blockchain.

In non-European regions, it is difficult to keep blocks synchronized. It is impossible to perform block synchronization in Asia. When the mainnet is launched, will you provide peers in Asia? And the current Sei blockchain is difficult to maintain using cloud infrastructure. This is because disk IO usage is large. Or Could I use Bare Metal Service?

In my case, i running 2 fullnode in aws EC2. AWS EBS has difficulty accommodating Sei blockchain's IO utilization. How could i solve this problem? Also Could you give me some advice to run Pruning options?

pruning = "custom"

pruning-keep-recent = "3000" <-- it look like 300 sec , pruning-keep-every = "0" <- i can't understand pruning-interval = "10" <- what's mean of interval ? it told me every each 10 block node will pruning?

thank you

[philipsu522]

hey @victor0807 , when the node does pruning, depending on how fast your disk is, it ends up potentially causing spikes in IO, and if tuned in correctly the IO could end up slowing your node. Let's try changing "pruning=default", since that will take snapshots less frequently. Another thing you could do is to raid your disks so you can parallelize disk io. I'll keep this open until you report back

[victor0807]

EC2 is a virtualized disk, so IO is relatively slow. So I purchased an m.2 disk and am testing it. Additionally, I try to synchronize newly installed nodes by changing to AWS Europe area, but I can't connect well because I don't have enough peers.

I'm an employee of the cryptocurrency exchange. Synchronization of nodes is paramount to maintain stable service. However, despite numerous tests, synchronization in Asia was impossible even with EC2 or Bare Metal.

I'll try to find a way more. Oh, by the way, can you give me a rough idea of the release schedule of the mainnet?

thank you

[victor0807]

A new node is being installed in the German region. This is what happens every time you install a new node. The following happens when state-syncing. Usually, if the peer is connected after the error, the fetch block and apply block processes are performed. But the peer doesn't connect well.

After that, the 'send message fail' issue repeats infinitely like the contents of the second picture. Have you ever been through this situation?

One more question. Look at this. As you know i run a node by default option. but can you see?

When i happen this situation, i usually restart node. It look work well. but few minutes later it happened again.

please give me solution to solve these problem. Thank you

[victor0807]

I found why state sync working bad, because peers were so small. I restart node more than 500 times and finally it connected with peers. It could be solve when mainnet launch.

[philipsu522]

Thanks @victor0807 , closing