[General] external node v28.7.0 alpha for another chain, failed to start

[donavernokg]

Team or Project

No response

ZK chain

Era

Environment

Mainnet

L2 block number

No response

Provide a brief description of the functionality you're trying to implement and the issue you are running into.

Hello I am working on another chain (Sophon) that uses zksync external node. I am also requesting help from the dev team, but I figured asking experts here would certainly be great too. Currently I am facing an issue when starting the node, with snapshot sync off, syncing from genesis.
v28.7.0-alpha error log.txt

All files are taken from the repository linked above. This is what my docker-compose is right now,
v28.7.0-alpha docker compose.txt

I realized that the generate secrets are generating with logs included which leads to an error, but that's not my main issue.

external-node-1     |     control characters are not allowed
external-node-1 exited with code 1
external-node-1     | Error: failed reading YAML map from "/configs/mainnet_consensus_secrets.yaml"

Before this, I was running v28.6.0 fine with snapshot recov. When I started a new node without snapshot recov, I received errors

Error: One or more tasks failed: ["Task consensus_fetcher failed: consensus actor: external node executor: Components for pregenesis stopped: Network stopped: server_addr.bind(): Address already in use (os error 98)"]
Error: One or more tasks failed: ["Task consensus_fetcher failed: consensus actor: external node executor: Components for epoch 0 stopped: Debug page server stopped: TcpListener::bind(): Address already in use (os error 98)"]

Hence now I am trying out v28.7.0-alpha without snapshot recov. I followed this and the next chapter.

I tried adding in "genesis_spec" to my mainnet_consensus_config.yaml

genesis_spec:
  chain_id: 50104 # chain id
  protocol_version: 1 # consensus protocol version
  validators:
    - key: validator:public:bls12_381:9882bafaade693500bc142affee109aab1e098348633c420a80b1ffec96739aecf29aae2dd46c297ce0e2824317a837310dbef955610a91d8df678ca6df5c128fa367803f764e3bb954f2fc5a53cbb58af0de2f6e4c9dd64fab2a1318fd0f7a3
      weight: 1
  leader: validator:public:bls12_381:9882bafaade693500bc142affee109aab1e098348633c420a80b1ffec96739aecf29aae2dd46c297ce0e2824317a837310dbef955610a91d8df678ca6df5c128fa367803f764e3bb954f2fc5a53cbb58af0de2f6e4c9dd64fab2a1318fd0f7a3
  seed_peers:
    - key: 'node:public:ed25519:33b37733427033ecaef793e0373f56ec4b9554697f9f34648b1d6918379b56d7' # node public key of the main node (copy this PUBLIC key from the secrets you generated)
      addr: '127.0.0.1:3054' # same as public_addr above

But overall, I was not able to get my node to start up. Would appreciate assistance, thank you!

Repo Link (Optional)

No response

Additional Details

No response

[ly0va]

Hello @donavernokg, let's address your original issue:

Error: One or more tasks failed: ["Task consensus_fetcher failed: consensus actor: external node executor: Components for pregenesis stopped: Network stopped: server_addr.bind(): Address already in use (os error 98)"]
Error: One or more tasks failed: ["Task consensus_fetcher failed: consensus actor: external node executor: Components for epoch 0 stopped: Debug page server stopped: TcpListener::bind(): Address already in use (os error 98)"]

These errors occur if you have a service running that is bound to some port, and then another service (like the EN component you're trying to run) tries to bind to the same port and fails.

• Sometimes this might occur when you restart the EN if you do that too quickly, e.g. if the services have not fully finished the clean up before the new instance is spun up.
• Or, if you're trying to run 2 different instances at the same time, make sure to edit the config so that no 2 ports are conflicting. To see the active services and ports they are bound to, try running ss -tulpn or equivalent command.

The issue with yaml,

external-node-1     |     control characters are not allowed
external-node-1 exited with code 1
external-node-1     | Error: failed reading YAML map from "/configs/mainnet_consensus_secrets.yaml"

Seems to be caused by a malformed yaml config. Please check your file has any control characters unsupported by the yaml specification. They can sometimes be caused by editing in some GUI editors.

Please feel free to unanswer this discussion if you have any further issues.

[donavernokg]

Hello, I did check the yaml config and solved the malformed issue. I did check for used ports (all ports used in docker-compose.yaml) before starting the node. There were no existing services using it.

After running the node and letting it restart everytime it fails to start-up. I checked and found this, there does not seem to be any conflicts.

tcp   LISTEN 0      65535        0.0.0.0:3000       0.0.0.0:*    users:(("docker-proxy",pid=3812505,fd=4))  
tcp   LISTEN 0      65535           [::]:3000          [::]:*    users:(("docker-proxy",pid=3812512,fd=4)) 
tcp   LISTEN 0      65535        0.0.0.0:3054       0.0.0.0:*    users:(("docker-proxy",pid=3818679,fd=4))               
tcp   LISTEN 0      65535           [::]:3054          [::]:*    users:(("docker-proxy",pid=3818686,fd=4)) 
tcp   LISTEN 0      65535        0.0.0.0:3060       0.0.0.0:*    users:(("docker-proxy",pid=3819263,fd=4))               
tcp   LISTEN 0      65535           [::]:3060          [::]:*    users:(("docker-proxy",pid=3819269,fd=4))
tcp   LISTEN 0      65535        0.0.0.0:3061       0.0.0.0:*    users:(("docker-proxy",pid=3819798,fd=4))               
tcp   LISTEN 0      65535           [::]:3061          [::]:*    users:(("docker-proxy",pid=3819804,fd=4))
tcp   LISTEN 0      65535        0.0.0.0:3081       0.0.0.0:*    users:(("docker-proxy",pid=3820354,fd=4))               
tcp   LISTEN 0      65535           [::]:3081          [::]:*    users:(("docker-proxy",pid=3820360,fd=4))
tcp   LISTEN 0      65535        0.0.0.0:5000       0.0.0.0:*    users:(("docker-proxy",pid=3821062,fd=4))               
tcp   LISTEN 0      65535           [::]:5000          [::]:*    users:(("docker-proxy",pid=3821068,fd=4))

[ly0va]

Please check configs/general.yaml for conflicting ports as well. I see 3054 port is used, which is the default port for the consensus component. This being occupied by a some docker container further confirms the suspicion that is it the old node instance, so be sure to clean old running docker containers.

[donavernokg]

Thanks for the swift response. Before starting the nodes, I checked that all ports are not used. The example above are ports that I checked after the start-up.

It seems to be an issue with my config having
debug_page_addr: '0.0.0.0:5000'

changing the ip/port does not have an effect, but removing this does fix my issue. Although I will have to figure out what it is used for. Thank you

[ly0va]

@donavernokg It's an address that exposes the node's debug page.

Glad your issue is resolved, we will look further into this config option to determine if this is a common issue.

Closing this for now.