One of the most important aspects of being a reliable Block Producer is to have a secure failover policy in place that prevents block misses, forks and other blockchain malfunctions.
The Master-Master architecture that we are intending to use provides a high level of flexibility, allowing just one data center to sign the blocks, while every data center is ready to sign simply by changing the signature key in the blockchain. The activation of a specific data center (or failover) is done by sending a transaction to the EOS.IO blockchain which changes the block producer key.
The need for failover is detected by running an "observer" script that monitors if the block producer account has lost any blocks. In case this event occurs, the script will automatically send a tx that changes the signing keys of the backup server and this will immediately switch data centers and start producing again. Now, in order to prevent the observer from monitoring a fork chain, the script also estimates the participation rate of the blockchain, this is done to make sure than more than 80% of the BPs are in the same chain and is a requisite for the switch to be activated.
This same approach has already proven to be very reliable as our team has used it to failover in BitShares and we encourage other Block Producers to implement similar protocols.
Social Media
eosargentina.io
steemit:
Twitter
MEETUP Group
Telegram
Youtube
