- We stabilized the system after Thursday's issues. The issues were caused by people wasting bandwidth. Spammers were making connections to our mailserver, and while the mail was easily identified as spam, they were still wasting bandwidth. Later, Javapipe successfully eliminated some attacks, and we also reinstituted the connection limit to make sure the system continued to run smoothly.
- Last night, I completed an analysis of how we can parallelize the system and determined that the task will be significantly less complicated than originally thought. This is excellent news. I think it will be feasible to accomplish it by December, not January or February as originally thought. The only issue holding us back is that we are finishing up our current commitments and can't devote 70-hour weeks to this until November 29.
- We hope to make a release this weekend with the first stages of the parallelization. We need to convert communication with the daemon servers to WAMP channels so that every copy of the server receives the same data. We can release this change and let it run for weeks, and do the same with a few other changes one at a time, reducing the number of possible bugs at the time we make the final release to increase capacity.
- Other features we'll be focusing on before Thanksgiving are tax preparation and transferring our own accounting methods and Coinbase payouts to use LTC as a base as much as possible. Now that it's clear that BTC will become unaffordable, we need to move quickly to consolidate our large wallets and eliminate as much usage of the network as possible before the fees get out of control.
- One of the benefits of the switchover to the enterprise-class connection is that the Microsoft E-Mail issues will be gone, because our system will be hosted on a different IP address.
- I closed a few tickets from some customers today about forks. Since Segwit2x is gone, that one is irrelevant now. With Bitcoin Gold, we do not plan to provide Bitcoin Gold or take steps to avoid replays of whatever Gold we have, because the daemon doesn't compile and its value is not high enough to justify the work and risks involved.
Discussion of development releases of Prohashing.
3 posts • Page 1 of 1
Last edited by Steve Sokolowski on Sat Nov 11, 2017 9:28 am, edited 1 time in total.
I thought 2x was back on ??
Is 11,000 workers still the theoretical limit for the current set up? It is currently at 10,238 workers and as it has increased from the high 9,000's over the weekend my efficiency has dropped as it has gone up, 97% down to 92-93%. I live in the UK.