A glitch occurred on 19th Aug 2016 which resulted in a superblock being created that only provided Neural Network statistics for Asteroids@home, Atlas@home and BURP.
The issue was resolved when the next superblock was staked with correct neural network stats, but whatever caused this glitch resulted in 3 projects being assigned 9.67 * more magnitude for a single day.
The potential consequences for this glitch are:
Daily earnings were affected (positively for those in 3 projects, negatively for the users in the other 26 projects).
DPOR stake weight will have shifted significantly across the network (Potentially affecting network consensus? If those unaffected are staking more frequently due to increased DPOR stake weight they may have a higher chance of accidentally repeating the glitch? This is all a guess though, I'm not certain how DPOR stake weight works 100%).
Voting weight was concentrated to a small quantity of users. 20% for a project before resulted in approx 2,606,896 magnitude weight, where as 20% after resulted in 25,200,000; considering several recent foundation expense polls only had approx 50mil voting weight participation, an individual with 25 mil voting weight from a glitch is scary. At the moment it's not a concern because we have human middleman separating the result of a poll from the whitelist/foundation, however if this was an automated voting system we'd have more to worry about.
When this glitch occurred I tried launching the Neural Network window, resulting in my client freezing - does launching the neural network window whilst the client is staking cause the client to freeze or could it be related to this issue instead?
After resetting the client, I ran 'execute syncdpor2' which resulted in all projects showing accurately, not just the 3 reported in the glitched superblock.
I could create a script for my grcnodes to query the current neural hash and the list of possible neural hashes - anyone know the exact commands for this? (I could create comparison tables to keep tabs on current superblock stats).
Initially reported by AEM7 on 19th Aug 2016
A glitch occurred on 19th Aug 2016 which resulted in a superblock being created that only provided Neural Network statistics for Asteroids@home, Atlas@home and BURP.
The issue was resolved when the next superblock was staked with correct neural network stats, but whatever caused this glitch resulted in 3 projects being assigned 9.67 * more magnitude for a single day.
The potential consequences for this glitch are:
When this glitch occurred I tried launching the Neural Network window, resulting in my client freezing - does launching the neural network window whilst the client is staking cause the client to freeze or could it be related to this issue instead?
After resetting the client, I ran 'execute syncdpor2' which resulted in all projects showing accurately, not just the 3 reported in the glitched superblock.
We discussed this issue last week in the 'Gridcoin Community Hangout #003'.
Recently, individuals have been reporting similar issues via the forum (0 mag, bad superblock again?): https://cryptocointalk.com/topic/1331-new-coin-launch-announcement-grc-gridcoin/page-1174#entry222581 https://cryptocointalk.com/topic/1331-new-coin-launch-announcement-grc-gridcoin/page-1174#entry222631
I could create a script for my grcnodes to query the current neural hash and the list of possible neural hashes - anyone know the exact commands for this? (I could create comparison tables to keep tabs on current superblock stats).
Cheers, CM.