Status as of Wednesday, May 24, 2017
Posted: Wed May 24, 2017 8:46 am
Here's a brief status update:
- Last night, Chris resolved one of the core issues causing instability with the system. He discovered that the database-operator, which stores share data and holds lookup tables in memory for customers' payout proportions, was running out of memory because we have had so many customers recently. For now, the simple solution was to increase the memory allocated to the mining server. In the future, I'll modify the tables so that they omit customers who haven't mined with the pool for a long time.
- See the post about the system upgrades that will be performed over the Memorial Day weekend. Upgrades are likely to begin tomorrow night and continue through Sunday night. There are so many components to install that it will take multiple actions to get them all finished.
- There are three additional stability improvements on the way. First, there is a failsafe being tested to restart the share inserter if it runs out of memory, so that even if something happens to it, mining will continue with maybe 5 seconds lost across the entire pool. Second, there are some general software improvements to record less data about rigs in the mining server's memory, which should lead to fewer processor cache misses and increased performance as a result. Finally, we'll be upgrading to newer hardware that should at least double mining server capacity.
- Michael was able to finish a few modifications to the payout threshold displays on the website to better clarify to customers when fees will be incurred. The new display makes it clear that increasing the payout threshold is free, and lowering it costs money.