My final answer (for windows) is using CGWatcher:  http://manotechnology.blogspot.com/p/cgwatcher.html

CGWatcher coupled with failover configurations in cgminer, will handle stratum disconnections, sick cards, drivers restarting, shares not being submitted, mhs decreases, and more.  Configure windows to start cgwatcher on startup, save cgminer settings to default conf, configure cgwatcher to start cgminer automatically when cgwatcher starts, check all montior boxes (except restart based on hours, and except start/stop by schedule, and let 'er rip. 

Cgminer poops out due to sick cards and network issues on my windows rigs, and cgwatcher restarts cgminer automatically.  My worst rig restarts cgminer about four times a day.  My best rig restarts cgminer about once every two days.

I too am impressed with DrHaribo's pool, efforts and presence.   It's been a stable environment, and truly "bitcoin mining made easy".  I'm pretty impressed with the miner client, which I tried because I couldn't get cgminer to run for days on end.  Although since I found cgwatcher, I'm back to cgminer.

Keep up the great work.

I'm using 7970s too, with stock voltage.  cgminer does the overclocking just fine, and I suppose it would do the voltage if the voltage wasn't locked on my gigabytes.  I wonder if TRIXX might install in Linux with mono/wine?  I'll try and post back.

To unlock the voltage, I'm considering updating the bios on the cards per this post:  https://bitcointalk.org/index.php?topic=186624.40

But for now, I just want the miners to keep working through the night.

I've concluded that Windows 7 Professional simply sucks.  I'm using cgminer 3.0.0 on Linux Mint, and it has been running for four days without issue.  My winblows machines are unreliable with cgminer 3.1.1, and have died many times in the same four days. 

My ISP briefly cuts internet access almost every night around 1:30am,  and cgminer doesn't like how winblows recovers.  It's just a hiccup for linux.  The bitminter client does better than cgminer on winblows when recovering from the interruption, but bitminter is not half the miner that cgminer is.

I'll be converting my winblows to linux.  So far, that's the only solution I'm able to come up with.

Thanks for the great tool and support, ckolivas.

I too watched cgminer rollover to my 50btc backup when bitminter went down.  (urls removed)


All OK, except I had round-robin set, and I would have expected cgminer to roll back to the primary pool at some time?  Perhaps round robin is only for the failover pools and does not include the primary pool?

Unfortunately, one card went 'sick', with cgminer throwing "Error -6", and continued to log this repeatedly every minute for 15 minutes until I restarted.  I don't know if it's the card, or the error is code related.  I restarted cgminer and the sick card sprang right into action. I include it here in case ckolivas might find it useful.


Thanks

FYI:  CK's wallets (both BTC and LTC) are in the readme's of his cgminer packages.  Scrypt readme for LTC, readme for BTC.

I had some initial issues with the 3.1.1 as recommended, but am diagnosing personal network issues.  I'll report back in a couple days, too.

drlukacs, I was going to start searching, as I know I've seen the instructions out here, but if you have a link or have some hints, I'd appreciate them.  I have Gigabyte GV-R797OC-3GD.  Thanks.

I'm inspired to, now!  Thanks.

I'm still struggling with "Stratum interrupted" issue. On windows, using CGMINER 3.1, I've updated the network card drivers to the latest, rewired all network connections, and have tried using a local stratum proxy.  My windows machines still fail at different times, throwing one "stratum interrupted" line in the log, and then cgminer does nothing.  I've seen others with multiple "stratum interrupted" lines in their logs, but mine throws one then seems to freeze.  CGMINER does not accept keyboard commands at this point.

In contrast, my linux rig, which is hooked up to the same network, manages to keep connected.

Is there a cgminer parameter I can try to make it reconnect?  Or at least make it's failover configuration work?  I believe I have the failover config correct, but it doesn't trigger with "stratum interrupted".  I might try going back to an older version of cgminer.

Thanks

drlukacs:

7970s have a 150 difference max between the engine and the memory.  So if your engine is at 1150, you can set your memory to 1000, but not 950.  At least on the gigabyte 7970s I have using cgminer.

Thanks, phillips.

I wonder if powered risers are as big of a necessity when the 7970 cards take 6 + 8 pin PCIE power drops from the PSU?  Is such a card also pulling lots from the PCIE slots and needs a powered riser? 

I've been assuming these issues were on single PCIE power drops from the PSU, not the 6+8 pins.

Thanks

I'm getting this, too.  Running 3.1 on windows, and 3.0 on linux.  Of three rigs, 1 windows and 1 linux both went "pool 0 interrupted" and ceased mining.  The third rig kept on mining.  Neither of the two that "interrupted" failed over.

I'll try setting up a local stratum proxy, but I'm not convinced it will help.

Thanks CK, cgminer is a terrific tool.  The new release is packed with updates!

We should have a sticky/faq for miners under 700mhs.  There are lots of repeat questions and chatter in the lower mhs ranks.  I'm only at ~2ghs and there's nothing to talk about from my perspective--fair, even, consistent and well supported pool...

I can get ~735-750mhs LTC with that gigabyte with the switches below (stock locked voltage).  LTC mining uses ~50w more per gpu than BTC mining.  The real edge is the memory to engine clock ratio.  Check cgminer's scrypt readme for good help finding the sweat spot.

--shaders 2048 --lookup-gap 2 --thread-concurrency 8192 -I 13 -g 2 -w 256 --gpu-engine 1100 --gpu-memclock 1500 --auto-fan

Stratum and getwork user counts no longer on the statistics page?   I suppose the hash rate for stratum and getwork load tells me the same thing....

I would like to know if there are other 'best practices' for the pool and forum.  Not listing what I would have thought was a public adr didn't cross my mind.  I'd appreciate learning any other no-no's.

Good effort, slush, thanks.  I wish there was a 'thanks' button for posts, so we could share the sentiments of other posts without bloating the lone, long thread.   Thanks.

As CK says of cgminer, the 7970 requires the memory clock to be no more than -150 than the engine clock.  Per the cgminer readme:

--gpu-memdiff -125
This setting will modify the memory speed whenever the GPU clock speed is
modified by --auto-gpu. In this example, it will set the memory speed to
be 125 Mhz lower than the GPU speed. This is useful for some cards like the
6970 which normally don't allow a bigger clock speed difference. The 6970 is
known to only allow -125, while the 7970 only allows -150.

Thanks for keeping a sharp eye, everyone.  I'm not currently paying such close attention to the blocks and rewards, since I'm assuming everything is working as planned.  It's nice to know there are people with their finger on the pulse, and a responsible sysop that is fixing things.