2026-06-22 Logos storage weekly

Week 25 2026 (Mon June 15 - Fri June 19, 2026) §

Highlights §

• After implementing several variationsof the Dynark paper idea, we have demonstrated it to be possible to run RLN proof times as low as 15 msecs; i.e., 13x faster than the original, and 4x faster than our previous optimizations.
• Our NAT traversal package is complete: this week, we have for the first time successfully transferred files over NATted nodes using TCP hole punching.

Status Integration §

• achieved:
  • PR feat: logos storage for History Archives status-go/pull/7486 - approved, but because the build system is quite flaky, we created another PR that we need to merge first so that 7486 reliably be merged:
    • PR fix: more robust clone targets for nim-sds and logos-storage native builds
    • it is very time consuming and the fact that the status-go team seems to be very busy, it is hard to align everything in place
• next:
  • we will still need a PR to bump the version of lib storage, but this does not have the highest priority right now.

Large Data Transport Layer for Mix §

• achieved:
  • chore: bump libp2p to v2.1.0
  • WIP draft spec
• next:
  • reference implementation for the state management

Hidden Services Over Mix §

• achieved:
  • Worked on a research post summarizing the draft Tor-style hidden service spec.
• next:
  • Finalize and publish the research post.
  • Research the Guard protocol over Mix, model the hidden service, and run simulation to understand the expected anonymity.

E2E Testing §

• achieved:
  • Diagnosed degraded logos.test bootstrap fleet issues: https://github.com/logos-storage/logos-storage-nim/issues/1455
  • Bootstrap monitoring: daily health check https://github.com/logos-storage/logos-storage-nim/pull/1460
• next:
  • After the logos.test fleet node configs were updated with the correct external ips, the network presets need to be updated with the new SPRs.

RLN Improvements §

• achieved:
  • implemented several variations of the Dynark paper idea in nim-groth16, with multiple improvements over the paper version
  • with these optimizations + some circuit optimization, achieved 15 msec RLN proof time, single-threaded (compared to the original 200 msec full Groth16); that is, achieving a 13x speedup vs. the original, and more than 4x speedup vs. the recently optimized prover
• next:
  • write a paper about the improvements
  • convince the Zerokit people about using this :)
  • port to Rust

Anonymous DHT Queries §

• achieved:
  • Hardened togglePrivateQueries: returns Result, refuses to enable when mix isn’t configured. Default-enable in start now skips when no DHT proxy is configured, so mix relay + DHT proxy-only nodes start cleanly.
  • Added cap for max connections and max inflight in mix relay dht. Also allowed it to run as a mix relay node only.
• next:
  • Test mix scaling performance in the DO cluster

Integration with Logos Core §

• achieved:
  • Update documentation and setup versioning
  • Improve the doctest and setup the hub test integration
  • Started to work on Mix feature integration
  • feat: expose storage metrics to openmetrics
    • feat: add metrics collector endpoint to libstorage
• next:
  • Continue to integrate Mix for Logos Storage Module and UI
  • Provide more doctest tutorials for Logos Storage Module and UI

NAT Traversal §

• achieved:
  • Complete manual testing and integration scenario
  • Apply first review suggestions
• next:
  • Wait for more review