Work is continuing rapidly on several fronts, first and foremost investigating the cause of download failures reported by the community. Our working assumption is that this is caused by chunks being replicated to nodes that are too far away from the close group so they cannot be reached when other replicas go offline. This obviously should not be happening at all, and we are checking our replication range algorithms to find out if they might have been affected by recent changes to the network. Weāve also developed some analysis tools to find out where chunks have been replicated to in order to test our theory. Itās only happening with a very small number of chunks, but one is one too many.
Elsewhere, @mick.vandijke and Lajos have successfully implemented paymaster for uploads in Dave (GUI front end). A paymaster is a mechanism that pays the gas fees on a userās behalf, meaning they donāt need to manage an ETH wallet, making for a smoother UX.
Plus, weāre making good progress with the Merkle tree transaction verification (a.k.a. gas fees minimisation) integration.
And donāt miss the latest video from @jimcollinson, in which he dusts off some old hardware found lurking in his cupboard and asks, āWhat if we could combine all that wasted capacity into one giant, encrypted hard drive that anyone could use?ā
What indeed. Comment, like and subscribe friends. You know the drill.
General progress
@anselme worked on A PR implementing Ant Merkle tree payments with a mock smart contract, defining client-to-node and client/node-to-smart contract interface types.
@chriso investigated the issue with misplaced records and some performance degradation. Internally, heās putting together a plan for the adoption of Service Level Objectives (SLOs) to improve monitoring, testing and user experience.
@mick.vandijke successfully implemented paymaster uploads in Dave, in collaboration with Lajos, and tested the MacOS signing process for all Autonomi binaries. Mick is now working on improving the UX for using the paymaster in Dave.
@qi_ma, @anselme and @roland are working on a tool to analyse where chunks are located. This includes being able to to check if nodes closest to a particular record are actually holding it. It includes kad::get_record checks and allows the user to probe quorum failures. Qi is now investigating the possibility that over time a chunk can move away from a close group, designing a testnet to verify any fixes.
And @shu has been digging deep into the logs to get a full picture of record placement and affects on performance. He and @roland are analysing quorum to ensure replication is happening as it should.
Finally, @vhongph implemented a cost calculation for selected files in Indelible, our network gateway for organisations.
Happy haunting! 

 
   )!
)! :
: Great to see.
 Great to see. website versions (such as Toast Blog) even from my laptop over mobile broadband.
 website versions (such as Toast Blog) even from my laptop over mobile broadband. ).
).
 
 




 
 

 auto-update
 auto-update  , only reason I havenāt install it for family and friends, yet I would always have to help update
, only reason I havenāt install it for family and friends, yet I would always have to help update 