thoughts on mmcfe+pushpool+stratum+litecoin software

[fcmatt]

Hey all.

I have been playing around with litecoin, pushpool, mmcfe, and stratum for a bit and it seems the software out
in the wild (on github) is not very reliable when put all together as a whole. Or maybe I should say mmcfe is the
weakest link.

Now I am not complaining here. Free software is great but after setting this all up and running litecoin in testnet
mode it appears mmcfe has some pretty lousy bugs. For example:

With only a single miner running at 260 k/hash with diff set at 16 so I do not create blocks super fast I notice
mmcfe fails to count every block when multiple fall within 1 minute. In other words the cronjob fails to work properly
if multiple blocks are found in one minute and the cronjob runs every 1 minute. So payouts do not work properly.

This can be easily seen at a mmcfe pool on the /blocksAuth page when the share column says "updating". The block
is always right next to another block which was found very quickly.

This makes me think many of us who mine litecoins at these types of pools are not getting paid properly. I cannot
even imagine what a feathercoin pool must be like as blocks are found so quickly compared to litecoin right now.

When I was in testnet mode I would get paid about every 5 blocks due to finding 5 blocks per minute on average.

So I suppose the operator of the pool is just keeping the reward?

Just throwing this out there as a lot of us trust these pools to pay us properly but somehow I doubt they are all
functioning properly.

If I decide to start a pool, as I was thinking about it, a lot of the software has to be redone. Stratum also unless
these pool operators found a source that has already been patched up.

Can anyone shed some light on this and give their thoughts? I do not think a perfect set of software should just
fall into my lap to create a pool but based on what is out there... i no longer trust most of these pools either especially
if they run mmcfe.

[thep33t]

Hey all.

I have been playing around with litecoin, pushpool, mmcfe, and stratum for a bit and it seems the software out
in the wild (on github) is not very reliable when put all together as a whole. Or maybe I should say mmcfe is the
weakest link.

Now I am not complaining here. Free software is great but after setting this all up and running litecoin in testnet
mode it appears mmcfe has some pretty lousy bugs. For example:

With only a single miner running at 260 k/hash with diff set at 16 so I do not create blocks super fast I notice
mmcfe fails to count every block when multiple fall within 1 minute. In other words the cronjob fails to work properly
if multiple blocks are found in one minute and the cronjob runs every 1 minute. So payouts do not work properly.

This can be easily seen at a mmcfe pool on the /blocksAuth page when the share column says "updating". The block
is always right next to another block which was found very quickly.

This makes me think many of us who mine litecoins at these types of pools are not getting paid properly. I cannot
even imagine what a feathercoin pool must be like as blocks are found so quickly compared to litecoin right now.

When I was in testnet mode I would get paid about every 5 blocks due to finding 5 blocks per minute on average.

So I suppose the operator of the pool is just keeping the reward?

Just throwing this out there as a lot of us trust these pools to pay us properly but somehow I doubt they are all
functioning properly.

If I decide to start a pool, as I was thinking about it, a lot of the software has to be redone. Stratum also unless
these pool operators found a source that has already been patched up.

Can anyone shed some light on this and give their thoughts? I do not think a perfect set of software should just
fall into my lap to create a pool but based on what is out there... i no longer trust most of these pools either especially
if they run mmcfe.

I made a bunch of tweaks and now 1 out of 100 dont get counted( automatically, and are counted via the script). Check the scripts folder, there is a script to run and payout those shares. But yes, it has to be run, or setup in a cronjob.

You can verify yourself by looking at the transactions page, but note, those paid out via the script will show up our of order.

If you dont mind answering, were you able to get stratum working at all (reliably or unreliably) with pushpool?

[fcmatt]

snip...

I made a bunch of tweaks and now 1 out of 100 dont get counted. Check the scripts folder, there is a script to run and payout those shares. But yes, it has to be run, or setup in a cronjob.

You can verify yourself by looking at the transactions page, but note, those paid out via the script will show up our of order.

If you dont mind answering, were you able to get stratum working at all (reliably or unreliably) with pushpool?

I did not notice that script. Thank you for pointing it out. I did not expect to have to run a script to fix such an
important aspect of a front end. Sadly I turned everything off for now and after running the script it did not fix
things like I expected. Just another thing to add to my list of things to review when I try again. I am going to guess
it needs to run at a certain interval also.

I was using viperaus's stratum fork from github. While it was "working" and accepting shares I never got to tweaking
how it put the data into mysql. I concentrated on pushpool and mmcfe at first.

So I would say no, it was not fully working yet. Besides getting data into mysql I would also have to verify it was
forwarding shares properly to litecoin (which i think it was but I need to verify for sure).

[thep33t]

I will look at that fork.In my experiments I was not actually pushing valid shares forward, so I never even worried about dumping them into sql, that part is easy

[fcmatt]

I will look at that fork.In my experiments I was not actually pushing valid shares forward, so I never even worried about dumping them into sql, that part is easy

I will try again tonight and report back.

[thep33t]

I will look at that fork.In my experiments I was not actually pushing valid shares forward, so I never even worried about dumping them into sql, that part is easy

I will try again tonight and report back.

looks like that branch has the same issue I had when I modified the code. Hopefully someone finds out what is up with that (a bunch of other people have stratum working...)

[fcmatt]

I will look at that fork.In my experiments I was not actually pushing valid shares forward, so I never even worried about dumping them into sql, that part is easy

I will try again tonight and report back.

looks like that branch has the same issue I had when I modified the code. Hopefully someone finds out what is up with that (a bunch of other people have stratum working...)

So what are people doing? One solves the issue and then spread it to friends without explaining the patch? Not sharing the secret sauce?

I messaged viperaus about this. Lets see what he knows otherwise i will work on it. Looks like viperaus has already put a lot of work into that fork.

[fcmatt]

I will look at that fork.In my experiments I was not actually pushing valid shares forward, so I never even worried about dumping them into sql, that part is easy

I will try again tonight and report back.

looks like that branch has the same issue I had when I modified the code. Hopefully someone finds out what is up with that (a bunch of other people have stratum working...)

Based on using the stratum fork I mentioned above with litecoin-testnet I see ./litecoind getinfo showing the blocks increase when
I run my stratum miner. So that is good.

mysql db:
shares_history has information
winning_shares has information

I will now let it run for about 200 greater then 16 diff shares. I dont want to hammer it to fast.

[fcmatt]

after letting it run for 207 blocks I conclude that stratum is working properly.
litecoin in test net mode sees the hashes being forwarded to it from stratum.
litecoins are put in the wallet. mmcfe without that script you mentioned works horribly. But with it is manages to limp along.
I need to verify that mmcfe actually functions right though.

For example: my miner is submitting shares with diff greater then 16. Every share should be a block solution.
So that means with 260 k/hash a share comes every few seconds. Say 20 seconds max.
So why do I see in mysql table "winning_shares" -> "share_count" some anomalies. The sheer majority are "1" which it should be.
But I see this also. a 0, a 3, a 5, and a 7 under "share_count". I have to wonder if I am submitting some stales, invalid, or working
on old data. I should have put cgminer into logging mode.

[thep33t]

after letting it run for 207 blocks I conclude that stratum is working properly.
litecoin in test net mode sees the hashes being forwarded to it from stratum.
litecoins are put in the wallet. mmcfe without that script you mentioned works horribly. But with it is manages to limp along.
I need to verify that mmcfe actually functions right though.

For example: my miner is submitting shares with diff greater then 16. Every share should be a block solution.
So that means with 260 k/hash a share comes every few seconds. Say 20 seconds max.
So why do I see in mysql table "winning_shares" -> "share_count" some anomalies. The sheer majority are "1" which it should be.
But I see this also. a 0, a 3, a 5, and a 7 under "share_count". I have to wonder if I am submitting some stales, invalid, or working
on old data. I should have put cgminer into logging mode.

yea, that version is working from what I can tell.. I can now compare his dumbed down version with mine to see what is up.

[fcmatt]

after letting it run for 207 blocks I conclude that stratum is working properly.
litecoin in test net mode sees the hashes being forwarded to it from stratum.
litecoins are put in the wallet. mmcfe without that script you mentioned works horribly. But with it is manages to limp along.
I need to verify that mmcfe actually functions right though.

For example: my miner is submitting shares with diff greater then 16. Every share should be a block solution.
So that means with 260 k/hash a share comes every few seconds. Say 20 seconds max.
So why do I see in mysql table "winning_shares" -> "share_count" some anomalies. The sheer majority are "1" which it should be.
But I see this also. a 0, a 3, a 5, and a 7 under "share_count". I have to wonder if I am submitting some stales, invalid, or working
on old data. I should have put cgminer into logging mode.

yea, that version is working from what I can tell.. I can now compare his dumbed down version with mine to see what is up.

what version are you using? slush? generalfault?
and i know what you mean by dumb downed but what specifically did you notice missing that makes you say that?

[thisway]

@fcmatt are you using pushpool for long pull (older miners etc.) and the stratum fork of viperaus for stratum support?

[thisway]

I just installed viperaus stratum version. If I run it it throws -> method getblocktemplate not found.
I just checked the API docu of litecoin. I can't find this method. Only getwork is referenced.

How did you guys made it work?

[fcmatt]

I just installed viperaus stratum version. If I run it it throws -> method getblocktemplate not found.
I just checked the API docu of litecoin. I can't find this method. Only getwork is referenced.

How did you guys made it work?

I just got back from vacation. Give me a day or two and i will find what i changed.
And yes, stratum and pushpool. My 4500 k/hash thrown at my new pool, sadly did not solve a block yet :-(

[thisway]

Hey fcmatt, thanks for your answer. I hope your vacation was restful.

I already found the problem with getblocktemplate. I just compiled the version from github. Seems to be much newer than the one on litecoin.org

You wrote:

Besides getting data into mysql I would also have to verify it was forwarding shares properly to litecoin

Have you been able to verify if shares are properly send to litecoin?

Give me a day or two and i will find what i changed.

I would really like to know the changes. Thanks!  

[fcmatt]

Hey fcmatt, thanks for your answer. I hope your vacation was restful.

I already found the problem with getblocktemplate. I just compiled the version from github. Seems to be much newer than the one on litecoin.org

You wrote:

Besides getting data into mysql I would also have to verify it was forwarding shares properly to litecoin

Have you been able to verify if shares are properly send to litecoin?

Give me a day or two and i will find what i changed.

I would really like to know the changes. Thanks!  

Litecoin-testnet in a box showed that stratum was working. I am now trying to prove it in the wild. Just wish i had double the hashing power to speed it up. No one really wants to mine at a pool that has not solved at least one block to show it functions.

[feeleep]

Hi,

Did you manage to get it work? I am really interested in launching stratum for my pool...

feeleep

[fcmatt]

Guys, i have not updated this thread because i have not found a block yet at my pool lite.coin-pool.com.
Once i find a block i will gladly share any and all code plus my setup to you. Help me out by throwing some hashing power my way. 20 ltc bounty to the person who solves the block.

[sheinsha]

Guys, i have not updated this thread because i have not found a block yet at my pool lite.coin-pool.com.
Once i find a block i will gladly share any and all code plus my setup to you. Help me out by throwing some hashing power my way. 20 ltc bounty to the person who solves the block.

Now that you have found a block, can you please share your currently working config and code?
Thanks in advance!

[fcmatt]

Yes. I can and should share it even though i did not make much in major change. I just need time. Imagine i just got in from tearing up my old driveway into a dumpster... By hand.

What exactly do ppl want? Mmcfe-litecoin, stratum, stratum-mining, pushpool? Just a diff or all the code minus my sensitive info? I am running centos 6.3