Stale and duplicate shares to be investigated tomorrow
Posted: Wed Sep 06, 2017 7:23 am
Hi,
We apologize for the issue with stale shares. Performance limitations are preventing users from being assigned coins in a timely fashion when a new block arrives, resulting in stale shares until the system can get round to sending the latest block.
We plan to spend four days investigating this issue as the top priority beginning tomorrow, and registrations remain closed to prevent the problem from becoming worse. In addition, we are working on ways to increase manpower within the next month to get ahead of these issues. Over the four days, even if we can't come up with a complete solution on day one, we should be able to release some performance improvements to reduce the number of stale shares somewhat, by reducing the number of coin switches (and therefore, overall profitability) but the lower number of stale shares would make up for it.
If Chris discovers that the system was inaccurately crediting miners for shares, then he will reimburse everyone going back as long as necessary. If it turns out that the shares were stale, which we think is almost certainly the case, then there unfortunately is no money with which to reimburse users.
Another issue to be investigated is one that hashingpro was able to reproduce consistently, where he said that certain user accounts always have difficulty connecting whereas others always work fine, with the same miners and the same location. This information will allow Chris to figure out what is wrong with those accounts.
We apologize for the issue with stale shares. Performance limitations are preventing users from being assigned coins in a timely fashion when a new block arrives, resulting in stale shares until the system can get round to sending the latest block.
We plan to spend four days investigating this issue as the top priority beginning tomorrow, and registrations remain closed to prevent the problem from becoming worse. In addition, we are working on ways to increase manpower within the next month to get ahead of these issues. Over the four days, even if we can't come up with a complete solution on day one, we should be able to release some performance improvements to reduce the number of stale shares somewhat, by reducing the number of coin switches (and therefore, overall profitability) but the lower number of stale shares would make up for it.
If Chris discovers that the system was inaccurately crediting miners for shares, then he will reimburse everyone going back as long as necessary. If it turns out that the shares were stale, which we think is almost certainly the case, then there unfortunately is no money with which to reimburse users.
Another issue to be investigated is one that hashingpro was able to reproduce consistently, where he said that certain user accounts always have difficulty connecting whereas others always work fine, with the same miners and the same location. This information will allow Chris to figure out what is wrong with those accounts.