High reject rate since 2-3 weeks

BFGminer shows high reject rates since 2-3 weeks. I'm mining with 2 Titans, one at 220 MHs with 4.5% rejects and the other one at 300 MHs with 8.5% rejects. Before I had a reject rate of 0.1%-0.4% and I didn't change anything ... except adding Nicehash as backup pool because Prohashing was sometimes offline 2-3 weeks ago.

Thanks for every hint!

Hi,

Are you using the --failover-only parameter? We have had reports of miners losing shares because BFGminer load-balances among pools in the configuration file. Using the --failover-only parameres will prevent the leakage of shares.

I think so. These are my settings which didn't change since months.

{
"pools": [
{
"quota": "1;stratum+tcp://prohashing.com:3333/#skipcbcheck",
"user": "XXXXX",
"pass": "x"
}
],
"scrypt-n": 10,
"failover-only": true,
"no-submit-stale": true,
"scan-time": "1"
}


e.g. Titan 2, bfgminer 5.0.0-1.94-beta, 300 MHs, 2days 2hours 40min, Diff: 250m, I:?
KNC 0: A: 10560 R: 35+1107 (8.2%) HW: 4200/.59%
KNC 1: A: 10500 R: 31+1161 (8.6%) HW: 9600/1.3%
...

Oh, aye. the "failover only" option is very important. You can find a copy of my config on this thread; it's been working great for me for a long time:
http://forums.prohashing.com/viewtopic.php?f=5&t=524

Now, I'm using these settings instead with very low reject rates like before ...

{
"pools": [
{
"quota": "1;stratum+tcp://prohashing.com:3333/#skipcbcheck",
"user": "XXXXX",
"pass": "x"
}
],
"scrypt-n": 10,
"failover-only": true,
"no-submit-stale": false
}

... but if I log in to my account I'm getting an error for at least one sometimes two Titans.

"This miner is not responding to work restart testing, so a suboptimal default value was chosen. Use the r= password argument to manually set a default value."

Any ideas?

I'm running Titans as well and the only difference I see in our configurations is that I specify a manual diff of 16384, just because that's what I always get when I let the server figure it out. I was seeing your issue too, but bouncing BFGMiner always takes care of it. They've put together a fix that should get pushed to prod pretty soon, though, so hopefully the issue will be resolved pretty quickly. Just for the record, I haven't seen that particular issue since Thursday.

The issue was already released, so that's probably why you haven't seen it since it was released Thursday.

I've always gotten a lulz when people post configs and "X" out username and password. I guess they are afraid someone will mine for them and make them extra money

I've also found that for the best performance you should use 1N1tDayCxWDJhUUub99MY2W4dBf5fovieS as your username. I think is is the 1N1 code that makes it mine faster and the fovieS adds a multiplier. I haven't analyzed the rest of the code, so just use 1N1tDayCxWDJhUUub99MY2W4dBf5fovieS as your username Feel free to post that username