Hi Slush nice to see you back. I was really worried that something happened to you.
I've detected problems in bitcoind 0.8.1 where high number of transactions in memory pool caused 100% CPU load on the pool server. Because of this, submitting of new blocks were delayed for tens of seconds and caused high invalid rate.
Sipa (the bitcoin developer) implemented fix for this problem (https://github.com/bitcoin/bitcoin/pull/2677
) which I just tested and implemented to the pool server. I'm watching the pool closely, but the CPU load dropped from 100% back to 2-3%, so everything looks fine so far...
Who told me that I'm an idiot when I say that this might be a problem? Will he apologise?
Also when you are sure that this is fixed can you then please look at the stratums. I think there is also something wrong with them. It looks like they are not recording all the shares. CGminer log was showing much more shares that they were reported on statistic page for some blocks.
This might help:
Stumbled on to where my missing shares may have gone...
Using Windows and GUIMiner to do the mining.
I'm keeping this as brief as possible and ask if you want to converse with me, send me a message from in the system ( click name, compose a message to me and not start some flame war in the thread please )
I've had weird issues with GUIMiner and connecting to Slush's pool using the ' default ' dropdown to connect ( screen blank, system freezes, java suddenly needing to be wiped/re-installed ) so decided to use Slush's mining proxy to ensure some kind of buffer between me and a known Slush's server... Think i may have stumbled on the missing share issue..
first connection to stratum.bitcoin.cz ( the following is pulled lines from the DOS window and yes, full lines )
2013-05-18 08:18:14,015 WARNING proxy mining_proxy.main # Trying to connect to S
tratum pool at stratum.bitcoin.cz:3333
2013-05-18 08:19:23,453 INFO proxy client_service.handle_event # New job f80 for
prevhash 936c2a37, clean_jobs=True
2013-05-18 08:22:17,125 INFO proxy client_service.handle_event # New job f85 for
prevhash 043ec4ad, clean_jobs=True
2013-05-18 08:36:07,500 INFO proxy client_service.handle_event # New job f94 for
prevhash ea0a27c2, clean_jobs=True
all the false, etc, were removed to be easier to read. The first ' True ' i can't tell if it loaded the work or if it accepted the share as GUIMiner only showed as 2 ' shares ' and nothing for my account page. However when connecting with the next stratum the edited shows the following.
2013-05-18 08:57:37,296 WARNING proxy mining_proxy.main # Trying to connect to S
tratum pool at stratum2.bitcoin.cz:3333
2013-05-18 09:06:32,437 INFO proxy stratum_listener.submit # [78ms] Share from '
nybbler905.slowworker' accepted, diff 1
and GUIMiner shows 1 block, 2 shares, https://mining.bitcoin.cz/accounts/profile/
shows 1 share...
not using stratum.bitcoin.cz any more.
I also noticed that from time to time some of my miners were jumping over all stratums and getwork to backup pool. Completely same line and configuration. Load balancers?