(Thanks to spackle for this writeup!)
The first week has been a wild ride, and there is a lot to share. Presented roughly by order of importance:
PR Improving Node Startup
Restarting a node while there is a large mempool can take over an hour in extreme conditions, since the node must process all the mempool transactions before resuming regular operation. 0xFFFC has addressed this with a massive speed up in PR 9376. These changes are now in use on the stressnet, to great effect.
Block Sync Size
Block propagation and network synchronization suffer at larger block sizes, which encourages the creation of alternative chains. Setting '--block-sync-size 1' allows nodes to handle larger blocks more easily, and this is now standard practice on the stressnet. Further updates are expected on this soon.
Ongoing Investigations
- Block sync size
- Wallet-to-daemon connectivity and performance enhancements
- Tx pool management (including flush_txpool processing time)
General Announcements
- Go to monitor.stressnet.net to see network stats
- explorer.stressnet.net is offline for now
- Node operators should upgrade to latest release for a smooth(er) stressnet experience. Older releases struggle with larger blocks, and ban too many nodes.
- Congratulations to strawberry, winner of the wallet draining competition
- Ongoing discussion is being held on Matrix at #monero-stressnet:monero.social and Libera Chat IRC on ##monero-stressnet.
Stressnet Stats
- This plot shows the block sizes on stressnet over the past week
- Largest Block: 2.49 MB at height 2521514
- Most transactions processed in an hour: 33074 (equivalent of 794000 tx per day)
Thanks for the update.