The core developers have identified a chain variable that we believe may alleviate some issues around transaction validation, block absorption and chain synchronization, and (potentially) help reduce write wear on SD cards. Unfortunately this may have a negative impact on the user experience of Hotspot owners looking at Explorer/App that we describe below.
The network needs action to get block times, beacon rate, and witness rates back to normal in the short term. We’re also currently seeing about 2.7k hotspots (across different maker types) who are struggling to keep up with sync again. This, combined with slow block times, is causing wide confusion and frustration in the Hotspot operator community so we plan on acting on this today.
We will be updating the witness_refresh_interval chain variable to 3000 blocks, or approximately 2 days assuming 60s blocks, (from its current value of 20000). We will not see the full impact of this change for about 2 days as witness refreshes are randomly spread out.
This variable is responsible for controlling how the gateway ledger stores information about past witnesses. This data was used actively in the past to construct multi-hop paths before HIP 15. It is no longer being directly used by the network but still incurs a heavy write overhead on the ledger. Since then, it has also been used to generate the recent witnesses list you see in the API, Explorer and App.
This may potentially create some confusing information in the recent witness list and counts that Hotspot owners see in the App and Explorer. There was already some existing confusion with witnesses > 5 days prior in this list. We plan to store and serve this data differently in the ETL service starting next week.
The core developers will be issuing this chain variable at 5:30 PM PT today. Thanks for your patience and support as we work through this phase of growth in the network.