|
It's not really a stability issue.. The downtimes this morning and yesterday morning were caused by me trying to get a little extra performance out of the site by changing the frontend to use mysql_pconnect instead of mysql_connect. Apparently these connections were not being cleaned up properly after being broken, which caused a connection leak of sorts between the web servers and the db servers. (It wasn't an actual mysql connection leak, but more of a certain app's tcp connection limit being reached.) I switched the webservers back to mysql_connect this morning, so that particular problem should not happen again.
The eventual fix is to move all the queries to either mysqli or PDO, as I suspect the connections were not being cleaned up properly by mysql_pconnect which, along with mysql_connect (and the entire php_mysql API) is deprecated and slated to be removed from php any minute now.
Tonight's outage was caused by something entirely different and baffling, as the web servers were all up, connected to the db, and serving pages fine, but the haproxy instance that balances the load between them was, for some reason, refusing to route the traffic.
|
|
|
|
@flound
do you plan to active CAP in the next time ?
EDIT
my question is not relevant.
CAP is disabled at cryptsy
They appear to have gotten most of the issues sorted out with the new client which is pending release (1.5.0). Once that is released and the network appears stable, I will activate CAP on the mutliport. But not until it's back on Cryptsy. |
|
|
|
By the way, I'd honestly like to help you solve your issues if you want to come back. I'm not trying to minimize or deny your issues and I do believe you. But just saying "multipool sucks because of X, that's why I left", is not very constructive or helpful.
Oh believe me I never thought you were and you have always been available for any questions I have had in the past. I don't think multipool sucks because of anything. It's an awesome service and pool. I just don't have the time to try and trouble shoot why my rigs work worse on this pool than others. It was a recurring problem, stratum disconnects that plagued me. The problem with trouble shooting it is that I am never just sitting in front of my miners when it happens and I have way to much to do to just sit there and wait for the problem to surface each day. I did find one reason for the disconnects a few days ago on the EU server and hopefully fixed it. The issue was stratum's checking of passwords, every time they would expire it would check every user's password, about 150-200, all at the same time. And it's single threaded so it blocks on every check. I thought I had pushed the updated DB code out to the EU server but apparently I hadn't. The updated code caches all of the passwords in one shot (a simple select all, duh). So that particular reason for the disconnects, at least, is gone. |
|
|
|
|
By the way, I'd honestly like to help you solve your issues if you want to come back. I'm not trying to minimize or deny your issues and I do believe you. But just saying "multipool sucks because of X, that's why I left", is not very constructive or helpful.
|
|
|
|
You said in that 18 hour window your rig was on we were mining LTC for almost half the time? It couldn't have had to many switches if that was the case. I know with my setup I have it set to change coins an check profitability every 5 mins. Even though I am switching almost every 5 mins I get a much better payout with the same hash power as on multipool. So the idea of switching coins often shouldn't affect profitability depending on the payout scheme of the pool. That said the more multipool switches the more stales/rejects I get. That's not the case when I use separate pools rather than one that port forwards to a different pool/port.
No I said "we" were mining it, i.e. the multiport. The point was it's a slow coin so there wouldn't have been a lot of new blocks to cause stales like if we were on WDC or something. However, the multiport switched coins about the normal amount of times. |
|
|
|
How are you getting 10%? 1667/(142216+1667) = 0.0116 or about 1.16% stales, not 10%.
Edit: Oh, you're talking about SS:. What do you have scantime set to? Try setting it to 30. Also try upgrading to 3.3.1.
Also I see you're using I:20. What happens when you decrease intensity to 13?
The only problem with I set to 13 is that you are gonna cut your hashrate down by a lot depending on your card, from my testing with my 7950's it cuts it down by almost 30-35% and the stale/reject rate goes from 3-5% to 2-3%. Not worth it at all. I've always gotten terrible hashrates with anything over 13. What cards do you have? 7970s |
|
|
|
I've just run an interesting 48 hour comparison. My mining rig runs at about 1460kh/s and I'd been dedicated mining CGB on multipool for about a week. I was getting about a 0.05 BTC return per day return on profit snapshot. Got a bit better when CGB went through the roof a couple of days ago. Exactly 2 days ago I switched back to the multiport for 24 hours and my return dropped to 0.026 BTC a day. Exactly 1 day ago I switched back to mining CGB and my return for the last 24 hours jumped back up to 0.045 BTC.
What makes this even more surprising is that the CGB exchange rate has dropped ~ 30% in the last 24 hours. I'm getting about 2 to 3% stales exclusively mining CGB compared to up to 20% stales with the shitcoins. LTC and NVC are about 1% stales.
Conclusion... Too much coin switching resulting in down time and too many shit coins with high stales result in lower profitability.
That was my problem with this pool as well. Way to many lost shares during pool switching, disconnections almost everyday, and a massive variance in day to day payouts. I run MultiMiner and use it to switch pools and have found it to be much more reliable in the long run than using multippol sadly. If the disconnects and the efficiency were more predictable I'd still be mining here. Was this on US server or EU? Sadly I have had this problem on both servers and now have finally given up on trying to get it to not disconnect everyday several times. The problem doesn't happen when we don't switch coins. Hence why your rig flound had very little SS and rejects, we only switched coins a few times in the time frame you showed. My rig that ran for over a month (screenshot on last page) has less than 1% SS and rejects. There were many switches in the past 18 hours, we switch to CGB on a very regular and predictable basis. In fact the screenshot with the connection to the EU server shows a switch without any share loss or even a stale. |
|
|
|
How are you getting 10%? 1667/(142216+1667) = 0.0116 or about 1.16% stales, not 10%.
Edit: Oh, you're talking about SS:. What do you have scantime set to? Try setting it to 30. Also try upgrading to 3.3.1.
Also I see you're using I:20. What happens when you decrease intensity to 13?
The only problem with I set to 13 is that you are gonna cut your hashrate down by a lot depending on your card, from my testing with my 7950's it cuts it down by almost 30-35% and the stale/reject rate goes from 3-5% to 2-3%. Not worth it at all. I've always gotten terrible hashrates with anything over 13. |
|
|
|
I stopped my other cgminer, but I had it running for about 18 hours on pool1.eu (6pm Pacific to 1pm Pacific), here's the result: (Northern California to Amsterdam). We were on Litecoin for almost half of that time, but we did mine WDC for quite a while also. About 0.9% stales.  |
|
|
|
|
How are you getting 10%? 1667/(142216+1667) = 0.0116 or about 1.16% stales, not 10%.
Edit: Oh, you're talking about SS:. What do you have scantime set to? Try setting it to 30. Also try upgrading to 3.3.1.
Also I see you're using I:20. What happens when you decrease intensity to 13?
|
|
|
|
This is my garage 2x7970 miner with 1 month uptime on pool1.us.
Efficiency: 99.39%
The speed is not as high as it could be, due to the summer weather I have it underclocked and undervolted.
How do you have so few stales? I have about 10% stale shares. Where are you? Northeastern US, using pool1.us. I usually see a couple of shares discarded when there's a new block, and occasionally a couple submitted and rejected with "Job not found". What's your miner setup? |
|
|
|
I've just run an interesting 48 hour comparison. My mining rig runs at about 1460kh/s and I'd been dedicated mining CGB on multipool for about a week. I was getting about a 0.05 BTC return per day return on profit snapshot. Got a bit better when CGB went through the roof a couple of days ago. Exactly 2 days ago I switched back to the multiport for 24 hours and my return dropped to 0.026 BTC a day. Exactly 1 day ago I switched back to mining CGB and my return for the last 24 hours jumped back up to 0.045 BTC.
What makes this even more surprising is that the CGB exchange rate has dropped ~ 30% in the last 24 hours. I'm getting about 2 to 3% stales exclusively mining CGB compared to up to 20% stales with the shitcoins. LTC and NVC are about 1% stales.
Conclusion... Too much coin switching resulting in down time and too many shit coins with high stales result in lower profitability.
That was my problem with this pool as well. Way to many lost shares during pool switching, disconnections almost everyday, and a massive variance in day to day payouts. I run MultiMiner and use it to switch pools and have found it to be much more reliable in the long run than using multippol sadly. If the disconnects and the efficiency were more predictable I'd still be mining here. I guess I should state again that I'm willing to work directly with anyone who's having a problem with stales to help you resolve the issue.. It helps both you and the pool if I can resolve these issues and improve everyone's efficiency. Please email me if you're having an issue with too many stales. |
|
|
|
I've just run an interesting 48 hour comparison. My mining rig runs at about 1460kh/s and I'd been dedicated mining CGB on multipool for about a week. I was getting about a 0.05 BTC return per day return on profit snapshot. Got a bit better when CGB went through the roof a couple of days ago. Exactly 2 days ago I switched back to the multiport for 24 hours and my return dropped to 0.026 BTC a day. Exactly 1 day ago I switched back to mining CGB and my return for the last 24 hours jumped back up to 0.045 BTC.
What makes this even more surprising is that the CGB exchange rate has dropped ~ 30% in the last 24 hours. I'm getting about 2 to 3% stales exclusively mining CGB compared to up to 20% stales with the shitcoins. LTC and NVC are about 1% stales.
Conclusion... Too much coin switching resulting in down time and too many shit coins with high stales result in lower profitability.
I think the problem is actually that we've been on LTC more the past couple days and have had relatively bad luck on it. There's also an issue with WDC when both us and middlecoin are on it that results in getting lower profitability, but there was recently a patch to WDC that will take effect on block 600,000 that will help this. |
|
|
|
|
Multipool is still on the same chain as the block crawler. I've been checking it every few hours.
|
|
|
|
This is my garage 2x7970 miner with 1 month uptime on pool1.us.
Efficiency: 99.39%
The speed is not as high as it could be, due to the summer weather I have it underclocked and undervolted.
How do you have so few stales? I have about 10% stale shares. Where are you? |
|
|
|
I've just run an interesting 48 hour comparison. My mining rig runs at about 1460kh/s and I'd been dedicated mining CGB on multipool for about a week. I was getting about a 0.05 BTC return per day return on profit snapshot. Got a bit better when CGB went through the roof a couple of days ago. Exactly 2 days ago I switched back to the multiport for 24 hours and my return dropped to 0.026 BTC a day. Exactly 1 day ago I switched back to mining CGB and my return for the last 24 hours jumped back up to 0.045 BTC.
What makes this even more surprising is that the CGB exchange rate has dropped ~ 30% in the last 24 hours. I'm getting about 2 to 3% stales exclusively mining CGB compared to up to 20% stales with the shitcoins. LTC and NVC are about 1% stales.
Conclusion... Too much coin switching resulting in down time and too many shit coins with high stales result in lower profitability.
That was my problem with this pool as well. Way to many lost shares during pool switching, disconnections almost everyday, and a massive variance in day to day payouts. I run MultiMiner and use it to switch pools and have found it to be much more reliable in the long run than using multippol sadly. If the disconnects and the efficiency were more predictable I'd still be mining here. Was this on US server or EU? |
|
|
|
|
Yeah, I think I'm going to change the way I'm connecting clients to MySQL, the way I'm doing it doesn't seem to be very reliable.
|
|
|
|
This is my garage 2x7970 miner with 1 month uptime on pool1.us. Efficiency: 99.39%  The speed is not as high as it could be, due to the summer weather I have it underclocked and undervolted. |
|
|
|
Hmm, bummer day. Been hashing with 5 MH/s for 11 hours, .042 on the profit snapshot  Actually, income has been really poor recently in general here. Mining straight LTC is better. Anyone else noticing this? It seems like it should be great, but... something like 14% rejects also. What server are you using? WDC has been tanking today, that could be part of it. Edit: CGB too. |
|
|
|
Can you check the script that calculates the unconfirmed and confirmed status of blocks. It hasn't run in over an hour on CGB.
And on the db rows, I'd suggest it out clear out the block data on everything older than 20 days or 1 months, that way it will stay manageable over the long haul.
It's running now. It gets behind whenever we mine WDC due to the finding of hundreds of blocks and the aforementioned huge confirmed rewards table. (moving from unconfirmed to confrimed takes a few seconds per block) CGB should run soon. |
|
|
|
|
Show Posts
