Thread by @radu_chis

Looking back at over one year of weekly tech reports is like looking down from the top of a mountain, after slowly & painfully climbing it, step by step. It’s a humbling & inspiring view. And so, we push forward. 1% better each day is all we need. Our week in #elrondtech 👇

1/ Situations, when forking can occur, have been reduced by improving the block data broadcast fallback scenarios, when validators in consensus group are responsible to send the cross-shard data in case the block producer fails the broadcast (intended or not).

2/ The antiflood mechanisms have had their parameters and functionality updated to match a real-world setup, thanks to the heterogenous #battleofnodes test network. The overall antiflood architecture is now fit for a large adversarial environment.

3/ The Peer Rating data observed in #battleofnodes has allowed us to better understand & adapt our algorithms. Adjustments are being made to ensure a fair validator rating system while maintaining blockchain liveliness as high as possible.

4/ Other notable achievements:
- Peer honesty measurement added to the consensus mechanism to handle specific consensus topic flood
- Added a p2p debugger that will be our window towards what really happens at the p2p level

5/ - Meta-chain is no longer stuck forever due to shards not including new valid meta-blocks after a meta-chain temporary freeze
- Both the originator and connected peers accepting an invalid message on the consensus topic are now blacklisted

6/ - Aggressive fuzz testing of the delegation smart contract unveiled a subtle bug that only manifested after ~300 random TXs
- API route /transaction/:hash/status now correctly returns the status of cross-shard TXs
- TX generation tool bugfixing & improvements

7/ - Elasticsearch database management improvements to ensure al TXs are saved with the correct status
- Added more integrity checks for Validators & TXs, such as UserName length, value length, etc
- Updates to docs.elrond.com

8/ - unJail transactions will now only work when sent from the node’s associated staking wallet
- Synching multiple tries for Start in Epoch is now faster due to requests always reaching peers that have the relevant data
- Added more test-case scenarios related to delegation SC

9/ - TXs are now first validated & saved before gossiped, reducing a host of invalid TX attacks
- Elrond Distributed Name Service & Sharded Smart Contracts integration complete
- Fixed a P2P close method deadlock issue that would leave a node in “limbo” after being shuffled out

10/ - Routing staking rewards to the system SC address no longer causes node panic at end of epoch; in such situations, the rewards are now distributed to the Protocol Treasury instead
- Max memory capacity for TX pools adjusted (lowered) via tests to estimate memory footprint

11/ - HealthService component added for debugging memory usage & diagnosable components such as the TX pools
- TX pools instrumentation improved with a deep diagnose procedure regularly called by the HealthService

12/ - Cache creation improvement via unique naming & max capacity tracking both individually & cumulatively
- Resumed work on node soft restart feature to allow better monitoring & graceful shard change when shuffled

13/ - Bugfix for the unjail transaction, that didn't restore validator in the "new" list

"Stay Hungry Stay Foolish" and tune in soon for more of the #Elrondtech ⚡️ which powers the ecosystem