The network operators went on to detail the processes necessary to fix the issue.
Next week the team will be releasing the autopsy results.
THORChain’s blockchain operators announced on Twitter that the network had crashed Thursday due to a software fault. Members of the team have said that the downtime is not related to financial stability and that they are trying to fix the problem as quickly as feasible.
A further update from the THORChain team was released some four hours after the first notification, explaining that the problem had been traced to non-deterministic factors at work between nodes.
The network operators went on to detail the processes necessary to fix the issue, such as identifying the cause of the non-determinism, publishing an update, and restarting the state machine, indicating that they were almost there.
Finally Fixed by the Team
Three hours later, the group reported that they had narrowed the problem down to a problem with string manipulation. The problem was traced down to code that was attempting to move the cosmos, which turned out to be nothing more than string manipulation.
The developers also noted that they failed to notice the problem in stagenet because the faulty message was switched out instantly and never written to the block. As a result, the mainnet was impacted since the faulty memo was written in the block and the swap to synth could not have occurred in the same block due to the queue.
As the fix was being sent out immediately, the team also requested that THORChain node operators wait for further instructions. The crew has just now (within the previous hour) begun tweeting about being back online. After a delay of almost 20.5 hours, the network is back up and functioning normally. Next week the team will be releasing the autopsy results.