Status as of Friday, June 9, 2017

[Steve Sokolowski]

Here's a brief update:

• We seem to have stabilized the system through temporarily increasing difficulties to reduce the number of shares being inserted.
• But some of the efforts have resulted in permanent improvements. Chris finally figured out the "open file descriptors" problem with WAMP - that turned out to be caused by too many connections, due to an explosion in API usage recently. The API as implementation on the website should also be more reliable now, keeping the live data from disappearing.
• Chris also figured out that another cause of the crashes in the middle of the night (after fixing the payouts, which had been the previous cause) was the vacuuming. There is an "nThreads" parameter, which was set to 2, and now he set it to 1. That seems to have stopped the server from getting behind, while still keeping the disks from overloading.
• Our goal this weekend will be to eliminate data like block hash of the block the miner is building upon, the IP of the user who submitted a share, and other data that is unnecessary to store in the database for shares. A block hash has takes up 64 bytes, whereas other data in a share row like double precision difficulty takes up only 8 bytes. Four weeks from now, once the data containing these columns has fallen off the back of the archive process, we will delete the columns that now contain only nulls and VACUUM FULL the table, reducing its size dramatically and allowing us to move it back to the faster disks. Plus, since we can ignore this data now, there is less CPU load writing it to memory and creating Strings, and after an initial risk there will be less stuff around to cause bugs. We'll be investigating what other data we can ignore and delete without reducing the featureset of the system.
• Chris corrected balances last night. There haven't been any interruptions so far today. If there are, Chris will correct balances again at midnight before payouts.
• Two factor authentication was delayed by these issues. The code is there but has a security vulnerability that is yet to be fixed. I'm not going to devote effort to fixing it and releasing it until we can do the data reduction mentioned above.
• We ask for patience with payout issues. We'll be glad to assist if it's been 24 hours since your payout was due. However, we don't guarantee a payout time of day because the bitcoin network is overloaded and unreliable. Our efforts to move the system to litecoin as a primary coin will result in more dependable payout times once the system has more capacity.

[GregoryGHarding]

Thanks for the update and keep up the good work guys!

[n00bminer]

Continuous Improvement is the way of life! Keep up the good work

[GregoryGHarding]

thanks for throwing in 24 hours of patience for payout. seems the majority of messages and posts in the morning are due to lack of payment and patience. people seem to believe because payout script runs at a specific time, they're gonna get paid at a specific time. im also curious if mike was awarded the ps4 or not due to sec issue :p

[Steve Sokolowski]

thanks for throwing in 24 hours of patience for payout. seems the majority of messages and posts in the morning are due to lack of payment and patience. people seem to believe because payout script runs at a specific time, they're gonna get paid at a specific time. im also curious if mike was awarded the ps4 or not due to sec issue :p

The security issue was my fault, so I gave him the PS4.

I gave him a list of 15 bullet points and he did all of the others successfully. In the instructions, I failed to tell him to change the "continue session" dialog box to require that the 2-factor code be entered there as well. Therefore, with the current version someone could theoretically walk up to a computer where someone was logged in 15 minutes ago and get back into the system. That's why I've held up the release of that.

[GregoryGHarding]

thanks for throwing in 24 hours of patience for payout. seems the majority of messages and posts in the morning are due to lack of payment and patience. people seem to believe because payout script runs at a specific time, they're gonna get paid at a specific time. im also curious if mike was awarded the ps4 or not due to sec issue :p

The security issue was my fault, so I gave him the PS4.

I gave him a list of 15 bullet points and he did all of the others successfully. In the instructions, I failed to tell him to change the "continue session" dialog box to require that the 2-factor code be entered there as well. Therefore, with the current version someone could theoretically walk up to a computer where someone was logged in 15 minutes ago and get back into the system. That's why I've held up the release of that.

in leu of 2FA activation is it worthwhile to incorporate a pin system for changing payout addresses? or will that just bring us up to the time it would take to fix the security issue with 2FA