##### Today's Update:As some of you may have noticed, we recently ran another successful 500 node HoloFuel stress test. This was great news as prior to this the issue with CPU load was preventing us from stress testing the DNA. We've finished up a round of integrated dev-test and are preparing to promote this release to a QA cycle where we run full regression tests on functionality.Here's what we have recently addressed in development to reach this point:- Holochain .48 fixed a problem where Sim2h assumed it could trust that the conductor would hold something when it sent a holding request. This was (of course) a bad assumption because the conductor could crash or some other error could pop up, (e.g.. it was missing some other data needed to do validation because of data arriving in different orders as is to be expected in distributed systems). The fix we deployed in .48 involved another communication round back to sim2h from the conductor acknowledging that it did indeed hold the data.- This fix increased computational and networking cost of validation as well as made each complete validation cycle take more time - which is why we began to see CPU load increases again.- The positive unintended consequence is that it also enabled us to see was a long-standing bug where a validation in progress (as opposed to waiting in the queue to be processed) was not being detected when a request arrived for validation, so the same validation was simultaneous being done many times. (With our 'fix' we saw as many as 25 simultaneous attempts at validation.)So in a round about way, the fix from .48 made that longstanding bug significantly worse, which was why we were failing final-exam tests. We've now fixed it and we've done a round of dev testing, and now we are proceeding cautiously, but hopefully.After we have done more aggressive manual testing with a wider QA team and we have confirmed that the Holochain point release we are using is ready for public use, we will proceed with a new blessed version of Holochain and expand testing of HP Admin and HoloFuel out to our core testers.
#### Today's Update:We completed the QA tests successfully yesterday and began adding new testers to the environment 1-by-1. This process was slow and careful. Our intent was to test the time that it takes each successive node to enter into an existing HoloFuel DHT space and sync with the DHT or their shard of the DHT. This has been an elusive behavioural test for us to do in a way that models the apps running for hosts on HPOS as opposed to the way we do in our final exam stress tests. The additional careful testing yesterday paid off and we were able to then add some optimisations to sim2h. That same issue regarding our fix to Holochain .48 also caused additional queuing in sim2h, which in turn led us to devise optimisations. The pattern here being: ensure validity, then optimise for performance.We just created a new testing environment this morning and we have moved quickly through regression testing to evaluate basic functionality. We are now proceeding with a small multi-node stress test, then we will additional host nodes with multi-node transactions to populate the DHT reasonably. After that, later today, we'll proceed with a 1-by-1 approach to adding Core Community testers into HP Admin and HoloFuel.