## EPF5 Dev Updates - Week 15

[◀️ *previous update*](https://hackmd.io/@hangleang/epf5-week14)

### Goals For This Week

- [x] dubug on peer disconnect issue
- [ ] reduce network overload on duplicate messages

### Activities

- finally, the [pull request](https://github.com/grandinetech/grandine/pull/34) got merged to `das` branch, which still need to do final review once ready and confirm no more bugs.
- switch to a new branch for debugging/fixing bugs, the well-known one is [negotiation timeout](https://github.com/hangleang/grandine/tree/fix/negotiation-timeout) which cause peer to disconnect due to unable to response to request.
- fixed an hidden issue in this [commit](https://github.com/hangleang/grandine/commit/f9abd77dfd59d8e5615fc92f65553b65ed21fedb) on `GET /eth/v1/node/identity` RPC response not follow the [spec](https://ethereum.github.io/beacon-APIs/#/Node/getNetworkIdentity)
- peer review on [`peerdas-metrics`](https://github.com/hangleang/grandine/tree/feat/peerdas-metrics), now created another pull request to `das` branch for the metrics
- interop test with other clients (2 lighthouse, 2 grandine, 1 supernode prysm; as prysm fullnode still have some issue)
*so far, so good. peers keep syncing with each other well (Synced 4/4) under config below*

participants:
  # Super nodes
  - cl_type: prysm
    cl_image: ethpandaops/prysm-beacon-chain:peerDAS
    cl_extra_params:
      - --subscribe-all-subnets
      - --minimum-peers-per-subnet=1
  - cl_type: lighthouse
    cl_image: ethpandaops/lighthouse:unstable
    cl_extra_params:
      - --subscribe-all-data-column-subnets
  - cl_type: grandine
    cl_image: ethpandaops/grandine:hangleang-fix-negotiation-timeout
    cl_extra_params:
      - --subscribe-all-data-column-subnets
  # Full nodes
  - cl_type: lighthouse
    cl_image: ethpandaops/lighthouse:unstable
  - cl_type: grandine
    cl_image: ethpandaops/grandine:hangleang-fix-negotiation-timeout
network_params:
  eip7594_fork_epoch: 0
  eip7594_fork_version: "0x50000038"
snooper_enabled: false
global_log_level: debug
additional_services:
  - dora
  - goomy_blob
dora_params:
  image: ethpandaops/dora:peerdas-as-deneb

another sneak peek round, with almost clients as fullnode, just put 2 clients as supernode
with following config:

participants:
  # Super nodes
  - cl_type: prysm
    cl_image: ethpandaops/prysm-beacon-chain:peerDAS
    cl_extra_params:
      - --subscribe-all-subnets
      - --minimum-peers-per-subnet=1
  - cl_type: lighthouse
    cl_image: ethpandaops/lighthouse:unstable
    cl_extra_params:
      - --subscribe-all-data-column-subnets
  # Full nodes
  - cl_type: lighthouse
    cl_image: ethpandaops/lighthouse:unstable
  - cl_type: teku
    cl_image: ethpandaops/teku:nashatyrev-das
  - cl_type: grandine
    cl_image: ethpandaops/grandine:hangleang-fix-negotiation-timeout
  - cl_type: nimbus
    cl_image: ethpandaops/nimbus-eth2:pdsync
    cl_extra_params:
      - --subscribe-all-subnets=false
      - --sync-light-client=no
network_params:
  eip7594_fork_epoch: 0
  eip7594_fork_version: "0x50000038"
snooper_enabled: false
global_log_level: debug
additional_services:
  - dora
  - goomy_blob
dora_params:
  image: ethpandaops/dora:peerdas-as-deneb

- no yet look into the pubsub duplicate messages

### Plan For Next Week

- debug on pubsub duplicate messages
- reconstruction data columns if more than half columns available

---
[▶️ *next update*](https://hackmd.io/@hangleang/epf5-week16)