[ANN][LTC][Pool][PPLNS][STRATUM] - ltc.kattare.com - burnside's Mining Pool

[KWH]

Definitely not acceptable.  I need to roll out some new hardware for the site.  Top basically looks like this all the time:

PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME COMMAND                                                                                                                                                                         
20690 pushpool  20   0 84104 9.9m 2140 R  100  0.3  59:09.78 pushpoold

100% CPU.  Ugh.

Any pushpoold experts have any ideas for ways to reduce the load?

If I get a quad-core or octa-core box, can I just spin up a bunch of pushpoold's pointed at the same db and memcached?

Cheers.

Sadly, I had to change pools, 47%+ stales were killing me.
Hope you can get it fixed soon.

[Michail1]

Same here.  Will come back though when fixed.  Config is saved.

The website also shows:   Server Load: low

[Rawted]

Definitely not acceptable.  I need to roll out some new hardware for the site.  Top basically looks like this all the time:

PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME COMMAND                                                                                                                                                                         
20690 pushpool  20   0 84104 9.9m 2140 R  100  0.3  59:09.78 pushpoold

100% CPU.  Ugh.

Any pushpoold experts have any ideas for ways to reduce the load?

If I get a quad-core or octa-core box, can I just spin up a bunch of pushpoold's pointed at the same db and memcached?

Cheers.

I've been mining ltc on your pool for a few days now and i think highly of it. I do get random periods of high stales as well. I would be glad to make a monetary donation to chip in on increased hardware. I plan on mining there for a long time. Let me know how i can help.

[burnside]

I've been watching it for a few days now.  I'm pretty sure it boils down to periodic spikes from the crons that process the backend data.

I have some new hardware in the pipeline, depending on my day job I should be able to get it going in the next couple of days.

Also, as tacotime mentioned, I may have to adjust the difficulty a bit.

Thanks everyone for being so patient.  I know it's not fun.

Cheers.

[burnside]

Ok, I'm going to be taking pushpool and the litecoind down in a few minutes to transition it to a different box.  It's not a great box (working on that!) but it should handle the load better if I split the services a bit.

I'll post again when I'm done and things are stable.

Cheers.

[burnside]

Ok, I'm going to be taking pushpool and the litecoind down in a few minutes to transition it to a different box.  It's not a great box (working on that!) but it should handle the load better if I split the services a bit.

I'll post again when I'm done and things are stable.

Cheers.

I'm done for now.  Pool efficiency seems to be climbing, and so far my stales are pretty low.

Note that the IP address has changed for the direct pool hostname.  If your mining software didn't pick it up, you might have to restart it.

Cheers.

[burnside]

Looking at it now a few days later, things seem a whole lot better.  Overall pool efficiency has improved significantly.

One of our users has also donated some hardware to us.  It's in transit now.  When it arrives we'll check it out, buy some drives for it (it's coming without drives) and hopefully it can take over all of the ltc.kattare.com mining pool services.  I'm stoked.  It should let us merge what is now 3 really old servers into one nice fairly-recent server. 

Cheers.

[burnside]

Quick heads up...

I have helped two users in the last couple of days who were trying to setup cpuminer and could not figure out why it was not working.  It was not exhibiting any errors, but their hashing was not showing up in their account.  Turns out that in both cases, when viewing the logs I could see their accesses to the server, but under username "TheEther".  No matter what settings they made, that username is always what comes through.

Thus it is my educated guess that wherever this cpuminer.exe (version 2.2.3) is being downloaded from, the version they've released has the username and password HARD CODED to always hash under username TheEther.

If anyone has any information as to where this hard coded compile of cpuminer is coming from, please let me know so I can update the thread and make people aware.

Cheers.

[g2x3k]

where did you download it i just checked github source + bins please post some links

[burnside]

where did you download it i just checked github source + bins please post some links

I've tried to get that answer from the second guy, with no success.

I did get this though:

Okay, it seems that whenever I do

--user=user
--pass=pass

it gives me that problem, command line or json file.

when i do
--userpass=user:pass

it works. Only tested on command line however. May be a bug?

I don't think it's likely to be a bug though.  Seems likely there are creds getting built into the .exe.

I will note that the first guy said he got it from the link on our site, which goes to: https://bitcointalk.org/index.php?topic=55038.0

If I download the 64-bit windows binary there, unzip it, and run strings on it I get:

Code:

~/tmp$ strings minerd.exe | grep Ether
TheEther.1:1

And if I 'strings minerd.exe | less' I can see:

Code:

cpuminer 2.2.3
Usage: minerd [OPTIONS]
Options:
  -a, --algo=ALGO       specify the algorithm to use
                          scrypt    scrypt(1024, 1, 1) (default)
                          sha256d   SHA-256d
  -o, --url=URL         URL of mining server (default: http://141.165.227.52:9332/)
  -O, --userpass=U:P    username:password pair for mining server
  -u, --user=USERNAME   username for mining server
  -p, --pass=PASSWORD   password for mining server
  -x, --proxy=[PROTOCOL://]HOST[:PORT]  connect through a proxy
  -t, --threads=N       number of miner threads (default: number of processors)
  -r, --retries=N       number of times to retry if a network call fails
                          (default: retry indefinitely)
  -R, --retry-pause=N   time to pause between retries, in seconds (default: 30)
  -T, --timeout=N       network timeout, in seconds (default: 270)
  -s, --scantime=N      upper bound on time spent scanning current work when
                          long polling is unavailable, in seconds (default: 5)
      --no-longpoll     disable X-Long-Polling support
  -q, --quiet           disable per-thread hashmeter output
  -D, --debug           enable debug output
  -P, --protocol-dump   verbose dump of protocol-level activities
      --benchmark       run in offline benchmark mode
  -c, --config=FILE     load a JSON-format configuration file
  -V, --version         display version information and exit
  -h, --help            display this help text and exit
algo
http://141.165.227.52:9332/
TheEther.1:1

So clearly the windows 64 binary has been tampered with.

Cheers.

[burnside]

Pooler is now aware of the issue.  Turns out it wasn't tampered with by anyone nefarious, he just accidentally posted a binary he uses internally.  He'll probably post a new one soon.

Cheers.

[tacotime]

Do you plan to get stratum? I'll bring my 2mhs back to your pool if you do

[burnside]

Do you plan to get stratum? I'll bring my 2mhs back to your pool if you do

I honestly haven't had much time to look at it, but yeah, I do plan on looking into it soon.

[matauc12]

Hey buddy, nice pool. Been on 24hours now because Pool-X crashed for 16hours which is somewhat unacceptable for me although I understand shit happens. 2% stale for now and everything is smooth on my end.

Now, I leave for a week so don't change the IP when im gone :p haha.

[burnside]

Hey buddy, nice pool. Been on 24hours now because Pool-X crashed for 16hours which is somewhat unacceptable for me although I understand shit happens. 2% stale for now and everything is smooth on my end.

Now, I leave for a week so don't change the IP when im gone :p haha.

heh, we'll try to hold down the fort while you're gone.

[Devious]

The stale and unknowns are WAY too high, what's going on?  I thought this was being addressed?  19% stale is not on my end, I've got 50MB fibre with a 32ms ping time, and I keep getting pool not providing work fast enough.  If this can't be addressed it makes no sense to stay. 

[2013-02-18 08:32:44] LONGPOLL from pool 0 detected new block
 [2013-02-18 08:32:47] Rejected 006bb902 Diff 608/31 GPU 0  (prevhash-stale)
 [2013-02-18 08:32:47] Rejected 0445a806 Diff 59/31 GPU 1  (prevhash-stale)
 [2013-02-18 08:33:02] Rejected 01f4d841 Diff 130/31 GPU 0  (unknown-work)
 [2013-02-18 08:33:03] Rejected 062b5f13 Diff 41/31 GPU 1  (unknown-work)
 [2013-02-18 08:33:05] Rejected 01e16d7e Diff 136/31 GPU 1  (unknown-work)
 [2013-02-18 08:33:08] Rejected 037ed51c Diff 73/31 GPU 0  (unknown-work)
 [2013-02-18 08:33:09] Rejected 048fa180 Diff 56/31 GPU 0  (unknown-work)
 [2013-02-18 08:33:10] Rejected 025d4668 Diff 108/31 GPU 1  (unknown-work)
 [2013-02-18 08:33:19] Rejected 03c5e4c7 Diff 67/31 GPU 1  (unknown-work)

[burnside]

The stale and unknowns are WAY too high, what's going on?  I thought this was being addressed?  19% stale is not on my end, I've got 50MB fibre with a 32ms ping time, and I keep getting pool not providing work fast enough.  If this can't be addressed it makes no sense to stay. 

[2013-02-18 08:32:44] LONGPOLL from pool 0 detected new block
 [2013-02-18 08:32:47] Rejected 006bb902 Diff 608/31 GPU 0  (prevhash-stale)
 [2013-02-18 08:32:47] Rejected 0445a806 Diff 59/31 GPU 1  (prevhash-stale)
 [2013-02-18 08:33:02] Rejected 01f4d841 Diff 130/31 GPU 0  (unknown-work)
 [2013-02-18 08:33:03] Rejected 062b5f13 Diff 41/31 GPU 1  (unknown-work)
 [2013-02-18 08:33:05] Rejected 01e16d7e Diff 136/31 GPU 1  (unknown-work)
 [2013-02-18 08:33:08] Rejected 037ed51c Diff 73/31 GPU 0  (unknown-work)
 [2013-02-18 08:33:09] Rejected 048fa180 Diff 56/31 GPU 0  (unknown-work)
 [2013-02-18 08:33:10] Rejected 025d4668 Diff 108/31 GPU 1  (unknown-work)
 [2013-02-18 08:33:19] Rejected 03c5e4c7 Diff 67/31 GPU 1  (unknown-work)

We do have some new hardware in the works.  It'll be a week or two before it's deployed.


This is what I see on the site right now:

Pool Hash Rate   80.716 MH/s
Pool Efficiency   96.20%

Few mins later, reloaded the pool stats page, looks like this right now:

Pool Hash Rate   81.845 MH/s
Pool Efficiency   98.05%

Seems pretty good.

A few minutes later:

Pool Hash Rate   81.492 MH/s
Pool Efficiency   80.57%

Ugh... but... Ah!  We also found the last block just a minute ago.  Maybe the drops are happening on the round boundaries.  That would explain the spikes in unknown-work?  If you're submitting work for a block that was just solved?  Your log seems to confirm that theory.  LP new block found, then all work rejected for a minute after that?

Cheers.

[ilich]

I get the exact same thing, right after a new block EVERY time (prevhash-stale and unknown-work). My average stales around 5-7%. Add the donation into account and loosing coins on this pool. I'm moving back to another pool.

[Tranz]

Hi, new here and to mining, but so far I like your site! Can't say much for the stales, as I am still pretty slow compared to the power users. But hopefully I get can things moving faster here in the future. Still learning and having fun!

Thanks and happy mining!

[burnside]

I get the exact same thing, right after a new block EVERY time (prevhash-stale and unknown-work). My average stales around 5-7%. Add the donation into account and loosing coins on this pool. I'm moving back to another pool.

If I understand it correctly, every pool has stales after a new block.  The only way to reduce that is to shorten the window a bit by using smaller shares.  (smaller work size)  I can't do that though because that causes higher load on the box.  I'll explore reducing it when we have the newer box in production.

The other thing to keep in mind is that the pool efficiency is essentially an inverted number of everyone's overall stales, so when I see this like I do right now:

Pool Hash Rate   86.485 MH/s
Pool Efficiency   98.22%

That means the pool right now has 1.78% stales.  If you balance that with the higher stales around new block time, I think for most it averages out to 2% - 3% per round.  (in my case, right now, 2.6%)

I could hide the new-block efficiency drops by calculating over a longer time period, or more shares, such that they'd get lost in the averages, but I actually prefer it like it is where we can see the swings more clearly.

Cheers.