Consider this the official start of the Great Autonomi Upgrade Campaign. To properly test changes to the way relay nodes work, we need at least 50% of nodes to be on the latest version (run
antnode –version
or check the Launchpad. It should say Autonomi Node v0.3.7
) If not, please update your nodes to the latest version and restart them.Currently, only about 45% of
antnode
s are version 0.3.7 or above, which is not enough for us to test the connectivity and rewarding of relay nodes. The latest version of the client, ant
version 0.3.8 also corrects some upload failures, so it’s worth updating that too. If you have antup
run antup update
from the CLI.
Sorting out uploads is currently the main focus of the team. We’re also looking at how we can ensure smooth backward compatibility so that we are not so dependent on folks upgrading.
There’s a hotfix on the way too, which contains improvements to the Launchpad UPnP, client operations and the Python API. So make sure you upgrade to that one once it’s ready, too.
Fist bumps
to @Traktion for his invaluable feedback on the archive/file API, and to @riddim for pointing out errors with the Python interface. Nice work guys.
General progress
Along with @mick.vandijke, @anselme has been digging into the relay rewards, where we are planning to reward relay nodes to recognise the greater burden they carry. He also worked with @bzee and Victor on Python bindings. The aim is to have the Python help(AnyTypeFromOurAPI)
working for all our API types.
@mick.vandijke added a GetRewardsAddress
query to the node that returns its reward address as a signed response. This is part of the optimisation steps to rewarding a relay node with a portion of the rewards received by its client.
@bzee added some missing class exports to the Python bindings in the API to fix an issue raised by @riddim.
@chriso has been working on a hotfix release to address a Launchpad UPnP connection mode error, an issue with payee notification, unnecessary client dialling tasks, the missing Python bindings, and a problem with the routing table that was hindering discovery.
Chris has also been keeping an eye, along with @qi_ma and @shu, on the node upgrade statistics. So far fewer than half the nodes on the network are on the latest version, which will have been the cause of a few problems. In addition, he set up a test environment with libp2p
trace logging enabled for the team to investigate the upload issues.
@dirvine updated the document scratchpad behaviour in the Python bindings.
Ermine is working on stability issues with the network spawner. When we perform churn operations with multiple node restarts, we see problems. Ermine is currently checking whether the churned nodes actually release the memory and smoothly exit , or simply run forever in background
.
Meanwhile, Lajos implemented changes to the payment vault smart contract and started rewriting unit tests for the payment vault, as there have been many recent changes to that. He also worked on improvements to the NFT claims contract, and communicated those to the auditors.
@qi_ma has been roadtesting the latest hotfix RC and looking at occasional chunk poof verification failures. Internal testing suggests things are much improved, but we need to test it at scale, and that means getting people to upgrade to the latest version in sufficient numbers (in case we haven’t mentioned it before ). He wrote a proposal for rewarding relay servers and nodes upgraded to the latest version with extra emissions. This is to encourage folks to run reliable nodes. Qi also raised a PR for an initial POC of a
smart_equation
, demonstrating the potential to have a non-block-chain-based smart contract.
@roland has been working on network upgradability, ensuring changes are backward compatible. Now that we’re live, we can’t just make changes and hope they work. We have to keep things stable and run backwards compatibility. Upcoming versions of Launchpad will include more information about packaged versions.
@rusty.spork has been testing out the Launchpad as well as badgering people to upgrade their nodes (Perhaps we mentioned that we really need you to upgrade? ).
@shu was on maintenance tasks for the influx db/dashboard components in our current infrastructure, and reviewing the monitoring infrastructure.
And Victor worked on the Python Client. The good news is Python now covers 90% of the Rust client, but testing tools are still lacking.