Medalla testnet is the final multi-client testnet before the Phase 0 launch of ETH 2.0. A bug took away most of the testnet’s validators offline and therefore it was halted on August 14, 2020.
When this happened one of the Six Medalla servers were affected. The service stated that the time of the day as being one ahead in to the future. Therefore, the system averaged out the misshape by recording the time on all of the servers ahead by 4 hours.
Since the validators incorrectly proposed the blocks and attested the future slots the bug affected everyone who were using the Prysm client. Thus a majority of users were affected. This risk however is now over.
It was expressed that the recent Medalla Testnet incident is a learning experience. The validator who stated this was Raul Jordan, Prysmatic Labs’ editor. He has also stated that the launch Time Table for ETH 2.0 stays on track.
Ethereum.Network tweeted: “ETH 2.0 Validator Says Medalla Testnet Incident Was a “Learning Experience.”
Sydney Ifergan, the crypto expert tweeted: “Ethereum (ETH) is already huge and it is not very weak to die. It is a powerful network and they have lot of people on board to get things going. They are going to rule when things fall in place.”
Ethereum (ETH) Evan Van Ness on 8 Reasons
Evan Van Ness talks about 8 reasons on why the Medalla testnet bug is bullish for ETH2. He has a YouTube Video on 8 reasons why the Medalla bug is bullish for Eth2. In his own words they are as under:
#1: The obvious: One unnecessary dependency has been removed from the Prysm client. No other team was relying on a third party for time stamping, and now Prysm is not either.
#2: Medalla never died! Despite 70% of validators immediately vanishing from the network and DOSing the chain with messed up blocks/attestations, the chain kept going. It didn’t die. Sure, it didn’t finalize. But this was a catastrophic event, and yet the chain kept going.
Medalla is finalizing again. Despite being a testnet where stakers have little incentive to keep their validators up, yesterday Medalla returned to justifying & finalizing Sure it took much longer than if there were real consequences, but it still started finalizing!
#4: It will now be easier for stakers to safely switch clients. Eth2 clients are now making it easier to export your validator’s database so that you can switch clients, if one client has a bug. Huge for individual validators; huge for the network: easier for chain to stay live.
#5: This underscored how important client diversity is. If no client has above 33% of the network, then not even a major bug in a client can halt the network from finalizing.
This means that there’s not even that much of a penalty getting knocked offline even if you stake with the #1 client (by validator share) and it has a major bug. Right now, while Medalla is live, you should try out a non-Prysm client
Reminder: YOU ARE INCENTIVIZED TO STAKE WITH A NON-MAJORITY CLIENT
#6 Individual stakers have a better idea if they should stake. If you aren’t prepared to update your nodes on relatively quick notice (especially if you run the major client), then staking may not be for you. You might also consider running Lighthouse, Lodestar, Nimbus or Teku.
Staking early on when eth2 launches will likely have a solid staking return, but there are commensurate risks.
#7 Client teams are now better prepared if this happens on mainnet. Individual client teams got to practice a fire drill for a catastrophic event. While it isn’t fun, it is real world practice for what can happen when the chain goes live.
#8 The clients are hardening themselves against catastrophic events. Because of the fire drill, client teams now know what will happen if ~70%+ of the network is serving them bad blocks and attestations.
They know where the bottlenecks are in their individual client and are busy streamlining and refactoring to improve resilience. Sync and peering will be 2-3x safer because of this incident.
Evan also expressed, Fin: In short, this is why we testnet. Everything went wrong and yet the chain stayed alive, and has returned to finalizing. That’s impressive resilience. We need and are incentivized for client diversity. That means *you*. One add on: Huge respect to @prylabs for doing such a great job getting people to stake on the testnet. The problem is they were too skilled at it!
Credit: Source link