Hey dev!
Can you please fix "-r 0" flag behavior? It's impossible to setup multialgo mining. I noticed there is also --exit-on-connection-lost flag which doesn't really do anything in my case (first connection).
Example with "-r 0 --exit-on-connection-lost -T 5" flags:
20181018 22:58:41 Starting on: mine.zpool.ca:3643
20181018 22:58:41 ApiServer: HTTP server started on 0.0.0.0:4072
20181018 22:58:41 ApiServer: Telnet server started on 0.0.0.0:5072
20181018 22:58:41 WARN: GPU #1: PNY GeForce GTX 1080 Ti, intensity set to 24, 16777216 cuda threads
20181018 22:58:46 WARN: Connection with pool timed out. Trying to reconnect...
20181018 22:58:52 WARN: Connection with pool timed out. Trying to reconnect...

CCminer behavior with "-r 0" flag is simple. At launch try connect to the pool. If not possible, exit and don't waste time (~1 second). If yes - start mining.
Once the pool drops the connection for the first time, exit right away.



Other closed source competitors seem to have "correct" behavior, but i'd like to stick with t-rex mostly because of attitude.

Thanks!

Sorry for arrogance. I uploaded BIOSes of my toasters and opened an issue 
https://github.com/OhGodACompany/OhGodAnETHlargementPill/issues/47

Debian/Ubuntu with 4.15 kernels, 390.48 drivers and cuda 9.1.

This isn't working on all 1080ti vendors, did testing on some rigs and can report not to work: Asus Turbo, PNY XLR8. So this tool is rather limited for large mixed rigs, but thanks anyway.

Otherwise, with "right" cards it works on any baremetal mining platform, up to 11*1080ti/rig, largest i've got here.
It does NOT work on paravirtualized systems on KVM and VT-d pci-e passthrough with mining guests, but these setups are very rare among conventional miners. Tried paravilrtualized Windoze also with same result.

Any chance of getting it to work with everything 1080ti? 

Please compile a linux version, you are loosing hundreds of cards with potential fees. Windows is useless for mostly anything, especially if you run 10-12 cards per rig. Also check different neoscrypt pool stats, majority are still on ccminer-tpruvot and ccminer-klaust.

This SA crap is screwing hundreds of Linux miners.
The nofee option is absurd, Claymore probably works 24/7 on his code optimizations and this is the reason for devfee. Otherwise what's the point to burn his brain cells to update mining software really FAST.

I'm sure majority of Linux users don't really care about source. Let's face it - we are stuck with binary blobs like fglrx/amdgpu-pro anyway. Personally i'm on Linux because of stability and management ease, pssh rocks. Also i happen to disagree with Microsoft policies and attitude. We all know that Windows may be very easily "cured" from licensing issues, however what's the point of using such a product if there are legal alternatives.
I guess i'll start to rebuild miner images with kvm and GPU-passthrough support to avoid similar situations...

Claymore, is there any chance to change the logic of pool failover?

Traditional failover (up/down) is kinda useless for ethereum pools. They usually are up, however find kadzillion excuses to reject shares.
I'm talking specifically about dwarfpool and ethermine. Time by time they catch a glitch and reject 100% of the shares.

The latest was dwarfpools "corrupt DAG". At the same time older GPUs on expanse dwarfpool with same miner (7.2) mined ok, so this was obviously issue on the pool side.
The pool is up, but 100% rejecting shares. This doesn't force the miner to switch to failover pool. As a result farm is not just generating rejected shares for half of the night, but also consumes full power for nothing.

Handling this situation miner-side is quite easy, 5 consecutive rejects = GTFO for 1 hour. I don't pretend that numbers 5 rejects and 1 hours are optimal, they are just suggestions.

There's a glitch with remote management in Linux, multiple miners from farm start dropping connections after some time (v4.5).
I have a nodejs webgui for monitoring (drowned somewhere in this thread), but i also tested official EthMan.exe via wine and via simple linux tools - it's loud and clear:

nc -zv m10.lev 3333
nc: connect to m10.lev port 3333 (tcp) failed: Connection refused

Restart the miner (no need to restart the machine):

nc -zv m10.lev 3333
Connection to m10.lev 3333 port [tcp/*] succeeded!

Miners are barebone 14.04 without iptables.
Please let me know if i need to provide additional info

Got it running finally!
http://p2plv.ditchyourip.com:51050/static/
Located in EU (Latvia). Fee is 0.5% plus 0.5% for p2pool dev.
Both router and server are on 1.5KVA UPS, uptime 5 months.
My local miners, scattered across a couple of garages in country, get 4-5% less rejects, compared to US and NZ nodes, thanks to low latency

Looks like a serious whale is solo mining.
Check http://explorer.unitus.info/blocks
There are so many single address rewards for blake, each reward to a new address most likely not to draw attention. If it was a pool , it wouldn't go unnoticed and there would be further movements from fresh mined coins.
Let's push p2pool with all we've got, i'm throwing almost 100GH at it 

The trick with removing bind moved me forward. However, there is now a more serious error in python "No module named blake_hash". Btw, i tried your pool code on other coin daemons, it works just fine. Trying on debian wheezy if it's relevant.

Full trace:
2015-01-07 19:44:13.865851 p2pool (version 13.3-379-g4c2764f)
2015-01-07 19:44:13.886790
2015-01-07 19:44:13.887006 Testing bitcoind RPC connection to 'http://127.0.0.1:50604/' with username 'unitusrpc'...
2015-01-07 19:44:13.922759     ...success!
2015-01-07 19:44:13.923082     Current block hash: 8fe37947a56ba44f7081934f5e6bd538d494c8877ee64e1c5f485532876dc218
2015-01-07 19:44:13.923299     Current block height: 13166
2015-01-07 19:44:13.923416
2015-01-07 19:44:13.923565 Testing bitcoind P2P connection to '127.0.0.1:50603'...
2015-01-07 19:44:14.029161     ...success!
2015-01-07 19:44:14.029857
2015-01-07 19:44:14.030344 Determining payout address...
2015-01-07 19:44:14.031202     ...success! Payout address: UVUR77QKPE67Fnjni42m3zbe1kwFoSuohS.
2015-01-07 19:44:14.031809
2015-01-07 19:44:14.032269 Loading shares...
2015-01-07 19:44:14.033028     ...done loading 0 shares (0 verified)!
2015-01-07 19:44:14.033636
2015-01-07 19:44:14.034122 Initializing work...
2015-01-07 19:44:14.168101 > Unhandled error in Deferred:
2015-01-07 19:44:14.208148 > Unhandled Error
2015-01-07 19:44:14.208473 > Traceback (most recent call last):
2015-01-07 19:44:14.208683 >   File "/usr/lib/python2.7/dist-packages/twisted/internet/defer.py", line 368, in callback
2015-01-07 19:44:14.208830 >     self._startRunCallbacks(result)
2015-01-07 19:44:14.208939 >   File "/usr/lib/python2.7/dist-packages/twisted/internet/defer.py", line 464, in _startRunCallbacks
2015-01-07 19:44:14.209099 >     self._runCallbacks()
2015-01-07 19:44:14.209303 >   File "/usr/lib/python2.7/dist-packages/twisted/internet/defer.py", line 551, in _runCallbacks
2015-01-07 19:44:14.209522 >     current.result = callback(current.result, *args, **kw)
2015-01-07 19:44:14.209634 >   File "/usr/lib/python2.7/dist-packages/twisted/internet/defer.py", line 1101, in gotResult
2015-01-07 19:44:14.209823 >     _inlineCallbacks(r, g, deferred)
2015-01-07 19:44:14.209953 > --- ---
2015-01-07 19:44:14.210110 >   File "/usr/lib/python2.7/dist-packages/twisted/internet/defer.py", line 1045, in _inlineCallbacks
2015-01-07 19:44:14.210222 >     result = g.send(result)
2015-01-07 19:44:14.210389 >   File "/home/trader/unipool/p2pool/node.py", line 218, in poll_header
2015-01-07 19:44:14.210605 >     handle_header((yield self.factory.conn.value.get_block_header(self.bitcoind_work.value['previous_block'])))
2015-01-07 19:44:14.210668 >   File "/home/trader/unipool/p2pool/node.py", line 200, in handle_header
2015-01-07 19:44:14.210789 >     if not (self.net.PARENT.POW_FUNC(bitcoin_data.block_header_type.pack(new_header)) <= self.bitcoind_work.value['bits'].target):
2015-01-07 19:44:14.211026 >   File "/home/trader/unipool/p2pool/bitcoin/networks/unitus_blake.py", line 20, in
2015-01-07 19:44:14.211368 >     POW_FUNC = lambda data: pack.IntType(256).unpack(__import__('blake_hash').getPoWHash(data))
2015-01-07 19:44:14.211548 > exceptions.ImportError: No module named blake_hash
2015-01-07 19:44:14.230779 > Error while requesting best block header:
2015-01-07 19:44:14.231077 > Traceback (most recent call last):
2015-01-07 19:44:14.231346 >   File "/usr/lib/python2.7/dist-packages/twisted/internet/defer.py", line 551, in _runCallbacks
2015-01-07 19:44:14.231602 >     current.result = callback(current.result, *args, **kw)
2015-01-07 19:44:14.231774 >   File "/usr/lib/python2.7/dist-packages/twisted/internet/defer.py", line 1101, in gotResult
2015-01-07 19:44:14.232020 >     _inlineCallbacks(r, g, deferred)
2015-01-07 19:44:14.232312 >   File "/usr/lib/python2.7/dist-packages/twisted/internet/defer.py", line 1043, in _inlineCallbacks
2015-01-07 19:44:14.232493 >     result = result.throwExceptionIntoGenerator(g)
2015-01-07 19:44:14.232704 >   File "/usr/lib/python2.7/dist-packages/twisted/python/failure.py", line 382, in throwExceptionIntoGenerator
2015-01-07 19:44:14.232963 >     return g.throw(self.type, self.value, self.tb)
2015-01-07 19:44:14.233206 > --- ---
2015-01-07 19:44:14.233540 >   File "/home/trader/unipool/p2pool/util/deferral.py", line 41, in f
2015-01-07 19:44:14.233699 >     result = yield func(*args, **kwargs)
2015-01-07 19:44:14.233863 >   File "/usr/lib/python2.7/dist-packages/twisted/internet/defer.py", line 1045, in _inlineCallbacks
2015-01-07 19:44:14.234136 >     result = g.send(result)
2015-01-07 19:44:14.234306 >   File "/home/trader/unipool/p2pool/node.py", line 218, in poll_header
2015-01-07 19:44:14.234547 >     handle_header((yield self.factory.conn.value.get_block_header(self.bitcoind_work.value['previous_block'])))
2015-01-07 19:44:14.234802 >   File "/home/trader/unipool/p2pool/node.py", line 200, in handle_header
2015-01-07 19:44:14.235039 >     if not (self.net.PARENT.POW_FUNC(bitcoin_data.block_header_type.pack(new_header)) <= self.bitcoind_work.value['bits'].target):
2015-01-07 19:44:14.235343 >   File "/home/trader/unipool/p2pool/bitcoin/networks/unitus_blake.py", line 20, in
2015-01-07 19:44:14.235505 >     POW_FUNC = lambda data: pack.IntType(256).unpack(__import__('blake_hash').getPoWHash(data))
2015-01-07 19:44:14.235760 > exceptions.ImportError: No module named blake_hash

I can't make the node public, i'm on 5 mbps DSL. I doubt it's enough, even though i have nice latency. The reason i want to put up a private node is cuz i'm in EU and have really bad pings to your overseas nodes.
Thanks

Can't get the p2pool node running.
Latest repo cloned https://github.com/nzsquirrell/p2pool-nzs

unitus.conf:
rpcuser=unitusrpc
rpcpassword=x
server=1
listen=1
logtimestamps=1
gen=0
daemon=1
bind=10.0.4.2
rpcallowip=10.0.4.1
rpcallowip=127.0.0.1
rpcport=50604
maxconnections=20

~/unitus_pool$ python run_p2pool.py --net unitus_bla
2015-01-06 21:00:04.270700 p2pool (version 13.3-379-g4c2764f)
2015-01-06 21:00:04.270907
2015-01-06 21:00:04.271104 Testing bitcoind RPC connection to 'http://127.0.0.1:50604/' with username 'unitusrpc'...
2015-01-06 21:00:04.301014     ...success!
2015-01-06 21:00:04.301278     Current block hash: bee05d475225005e2802200539e39f5bc031d286ef0667dc83ebf4c38690fc93
2015-01-06 21:00:04.301424     Current block height: 11999
2015-01-06 21:00:04.301512
2015-01-06 21:00:04.301628 Testing bitcoind P2P connection to '127.0.0.1:50603'...
2015-01-06 21:00:09.302095     ...taking a while. Common reasons for this include all of bitcoind's connection slots being used...

It gets stuck @ p2p connection every time. I tried starting p2pool right after unitusd to make sure there are available slots, nothing changes.
Any ideas? Thanks

Yeah, you are right, Elpida memory... I found other 7970 cards with Hynix, they do much better

Thanks for Linux miner!

OS Ubuntu 14.04, catalyst 14.4, GPU Miner v7.0 Beta catalyst 14.6
Host 1: Sempron 140 2GB RAM, dual R9 290 @ 947MHz 1280h/s
Host 2: Xeon E5-2665 4GB RAM, dual R 290 @ 1040MHz, 1390h/s
Host 3: Celeron G1820 4GB RAM, quad 7970 @ 925 MHz, 1420h/s
Host 4: Dual Xeon E5-2665 16GB RAM, quad 7970 @ 925 MHz, 1420h/s

Flying good for the last 24 hours, no crashes.
I don't have any issues regarding virtual memory. Actually i don't have any virtual mem since Ubuntu is PXE booted into RAM (squashfs image) and uses approx 1.1 gigs, which leaves 900 megs free for the first machine.
Are hashrates normal for these cards?
The only thing that stops me from booting the entire farm with this miner is no way to monitor it. With c(s)gminer it's easy using the API and anubis. Any way to monitor Claymore's miner?

I found that the best way to switch pools is through cgminer API. If you have your GPUs highly overclocked, they don't really want to start every cgminer restart. With API there is no need to restart cgminer instance, unless you are switching between scrypt and sha256.
An example of adding a pool:
echo -n "addpool|stratum+tcp://stratum.give-me-ftc.com:3333,mylogin,mypass" | nc 127.0.0.1 4028

This is exactly why ASICs are bad. I realize that half of wholesale AMD vendors are retarded too, but there are ALOT of them to choose from.
Doesn't matter if these forums are good or bad, any sane company shouldn't behave like this. And if the employer doesn't send this forum troll to hell, they are just as bad as he is.
Most likely rumors about wtflabs burning rigs before shipping is also true after this. Monopoly's just suck.

P.S. Globally not happy about ASICs, anyways i'm happy i dealt with ngzhang and co.

Nice workaround! This stratum behavior is strange tho.
About active zero pool, atm i am trying: add new pools, switch new pools to highest priority, disable and remove old ones (with delays). It is working.
I was thinking of adding all of the pools to default cgminer config and disable them at startup. Bring em up when necessary (enable/increase priority). This should circumvent any add/remove glitches.

Same problem here. My workaround so far is:
clusterssh to the miners
echo "{\"command\":\"quit\"}" | nc 127.0.0.1 4028
killall -9 lxterminal
export DISPLAY=:0.0
sudo nohup lxterminal --command "sh /root/cgminer.sh"

cgminer.sh looks approximately like this:
cd /root
myip=$(/sbin/ifconfig|grep inet|head -1|sed 's/\:/ /'|awk '{print $3}')
wget -N --ftp-user=miner ftp:// minersrv.asd/$myip/cgminer.conf
cd /root/cgminer
export GPU_MAX_ALLOC_PERCENT=100
export GPU_USE_SYNC_OBJECTS=1
./cgminer --config /root/cgminer.conf

So, miners just massively pull configs from the server (defined by ip). This is kinda dirty since it involves restarting cgminer.
It's the only way if algorithm change is required (scrypt->sha256).
However, it would be really nice to add pools through the API for scrypt alts.

Anyone tried top of the line AMD APUs with this? Something like A10-5800K

If anyone is interested in an automated anubis install script:
https://sourceforge.net/projects/anubisauto/files/?source=navbar
This should work on any aptitude capable distro, it does install apache and mysql and sets right users/passwords. It uses P_Sheps git to get fresh version of anubis.

I was thinking of making a small linux distro, probably ~500 megs with anubis and cacti cgminer template from JinTu (https://bitcointalk.org/index.php?topic=76502.0) for peeps without linux knowledge, to boot from thumb drives. Anyone interested?

compile from source to stay out of trouble 
https://github.com/pocopoco/yacoin