roadmap/content/vac/updates/2024-04-02.md

6.9 KiB

title tags date lastmod draft
2024-04-02 Vac weekly
vac-updates
2024-04-02 2024-04-02 false

Vac 2024/04/02

vac:p2p:

vac:tke:

  • waku:general-incentives
    • Possibly continuing marketplace discussion with Waku (Martin)
  • waku:rln-membership:
    • Working on the proposal for RLN design (Martin)
  • status:SNT-staking
    • Supporting the SC team ad hoc (Martin)
    • Discussing using the staking contract at the org level (Martin)
  • status:L2-deployment
    • Further research into airdrop and incentive strategies of existing L2s (Martin)
  • nomos:mixnet-incentives
    • adjusted pricing function to account for measurement costs (Frederico)
    • verified that the modifications of the reward split scheme are correct (Frederico)
  • nomos:cryptarchia-wealth-concentration-estimated-stake
    • reviewed blog posts (Frederico)
  • codex:cdx
    • reviewed latest marketplace proposal (Fred

vac:dst:

  • eng-10ktool:vac:bandwidth-test
  • eng-10ktool:vac:bandwidth-test
    • Many many fixes to get Kubernetes with OpenvSwitch + offloading + VMs working
      • Reinstalled 3 nodes with new Debian + Proxmox flavour
      • Installed Mellanox OFED drivers
      • Experimented with VirtIO network, managed to eventually get SR-IOV and Virtual Functions working
    • Waku - Benchmarked 1-worker (one worker as one eighth of a 64 core node) cluster
      • Indications are we can scale to ~14k nodes if scaling is linear, vs CPU usage observed on 1-worker
      • Had 243 Waku nodes, including publishing, running on the worker or 1/8th node with headroom to spare
      • Network offloading appears to about 2x as efficient CPU wise when running Waku
    • Further fixes for offloading setup once SR-IOV was working
    • Waku - Reinstalled 24 workers, then wiped them all and reinstalled 8 of them :(
    • Diagnosed incredibly complicated packet loss issues (which turned out to be caused by cloned VMs - note to self - clean up /etc/machine-id next time!)
    • Waku - Benchmarked 8-worker cluster (1 physical 64-core), scaled to 1200 nodes, hit major issue with Calico
    • Added caching to Harbor, further investigated removing Harbor rate limits
    • Discovered that adding multiple jobservice workers to Harbor makes rate limits higher
      • Deployed 6 jobservice workers in Harbor
    • Removed Vaxis and Nia from Kubernetes to help with CPU accounting since they host worker VMs

vac:qa:

vac:acz:

  • secure-channels:waku:ethereum-chat
    • Finish the examination DCGKA ref implementation repo
    • Started to write a report about the examination of vector clocks used in DCGKA ref implementation
    • Checked that there is the motivation why we chose DCGKA in rfc
  • zerokit:vac:maintenance
    • github removed semaphore commit we used, was fixing CI issue

vac:sc::

vac:rfc:

vac:dr:

  • unstructured-p2p-improvements-survey
    • Looked into different aspects of libp2p specifications (including gossipsub versions and corresponding discussions). Also looked into the corresponding nim-libp2p works.
    • Followed discussions/PRs on libp2p specs and libp2p implementations

vac:nes:

  • state-separation:vac:state-separation-doc
    • Refined the State Separation PDAGs doc and add changes together with Ugur (Moudy + Ugur)
    • Worked on gathering important components for state separation (Moudy)
    • Researched and identified accumulators/nullifiers to integrate (Moudy)
    • Discussed monitoring with Moudy, and continued with monitoring (Marvin + Moudy)
    • Discussed with Moudy about PDAG report and next version of proposal on state-separation (Ugur + Moudy)
    • Started to write a draft of the next version of proposal on state-separation (Ugur)
    • Read about mutator set including Merkle Mountain Range and Bloom filters (Ugur)
  • proofsystems:vac:research-existing-proof-systems
  • proofsystems:vac:benchmarks
    • Kept working on the paper since new findings are arising (i.e. Nova Scotia not using Groth16) (Moudy)
    • Conducted server testing (Rostyslav)