More progress on the release process where we are now confident of being able to push out new releases on a bi-weekly cadence. This is primarily thanks to @chriso’s automaton skills and @shu building in metrics and monitoring on the staging releases, allowing us to more easily compare between them. It’s hard to fully explain this advance if you’re not in ops, but it’s a pretty big deal in terms of giving us a solid platform moving forward.
The sharp-eyed will have noticed a release candidate (2024.08.2.2) which contains some bug fixes and improvements, including showing download times and tracking bad node count. You’re welcome to try it locally, but we’re still working on eliminating the zero earnings bug for nodes from home, which will hopefully be done for the next release.
Talking of earnings, some lucky Beta testers hit the jackpot, becoming nano millionaires overnight. Sorry to say, this is most definitely not a feature: it is absolutely not our intention to create a digital dark ages where the super-rich few lord it over the rest of us zero-nano peasantry.
The disparity in earnings is part of the same root issues: connection troubles resulting in routing table that is not fully populated, leading to a calculated responsible range covering a wider (or even the entire) address space when nodes rejoin. This then results in a massively inflated quote.
Thankfully @qi_ma has some fixes in place for the up-coming release, and we’ll be adjusting the bonus nanos ratio to ensure the leaderboards don’t become skewed as a result.
The Launchpad is growing ever more useful and functional (some might even say beautiful), as Jim’s post here will verify.
More good news, this week Joab joined as our new finance manager. Welcome to the team Joab!
Thanks to @happybeing for the issue he raised on API controls for pricing discrepancies. @qi_ma is on this now. Also thanks to @loziniak for raising a PR and starting a thread on reusing a connection.
General progress
@roland is busy simulating the issues experienced by nodes-from-home users and is still working on a reliable fix for the zero nanos problem. We’re hopeful it will be ready by the next release, which is pencilled for next Tuesday.
@bzee is one of several team members who are debugging the ‘zero store cost’ issue on home nodes in the last few testnets. The cause seems to be that the home nodes are not reliably connected to the network (i.e. the rest of the network does not seem to be talking to the home nodes / have them in routing tables). This has yet to be confirmed, which is what @roland is digging into. @bzee also started refactoring some features of the client API.
@ansleme has been documenting the spend process and also looking at a replication bug where data is not always merged properly.
Also on documentation (vitally important as we near prime time), @jimcollinson is working on combining all our various docs together to create a resource for devs and users alike, and @rusty.spork has been chatting with folks who are interested in building on Autonomi, recommending resources and reading materials.
Jim is also working with @mazzi, who is building more features into the Launchpad, including the ability to select drives for storage and also ports and connection type.
@qi_ma continues to confer with the devs at libp2p
over range-based searching (Sybil resistance), and he’s also been looking into the data replication issue. Plus, he investigated community reports of high quote levels, raised a PR to add better tracking, and another PR to reduce the replication range.
@shu continues to add metrics so we can compare between test releases.
And together with Shu, @chriso has been working on multiple aspects of the release process to ensure we can maintain a standardised release cadence every two weeks, with the minimum of hassle. This is a “painting the Forth Bridge” type of job (i.e. never ending), but we’re definitely moving in the right direction.