[4+ EH] Slush Pool (slushpool.com); Overt AsicBoost; World First Mining Pool

[slush]

Sorry for the "me too!" response, but I also have >80-90% "invalid or stale" on two different computers and two different GPU's (nvidia and ATI) using m0mchil's miner -

this started happening an hour and a half ago, coinciding with slush's update post.  (in case that helps, or is just a terrible coincidence)  

Did some behavior change with this update?  Or is this not related at all?  I don't know enough more, just wanted to throw this out there just in case something is wrong.

I did not changed anything in core today, only added few stats views on existing data.

From ~300 active workers, I see that only few of them has a problem, so I don't expect it's on pool side. There can be a lot of possible troubles on miner configuration...

Users with problematic workers: vsladkov.worker1, vsladkov.worker0, everpex.gpu1, Tek022.Zack2, jakub300.gpu1, fabianhjr.x1-5970 That's all. Another 318 workers is going well. Please check your configuration, overclocking and so. Don't forget you have to use separate worker accounts for every software instance!

[zoidial]

Thanks for the info. 

I only mention because it was running fine for weeks and I haven't changed anything, so it's just peculiar timing to have errors now.  Just to clear it up, I have nothing overclocked and have separate worker accounts for each software instance.  About 1 in 15 is "accepted" and the rest are "invalid or stale".  I've restarted m0mchil's miner on both systems, but the results are similar.  I'll just wait it out if there's something odd happening.

I also wanted to mention that the updates are great, and thanks for this service, it's incredibly useful!

[slush]

I only mention because it was running fine for weeks and I haven't changed anything,

m0mchil fixed important issue on 5.January, so if you really 'havent changed anything', THIS is a problem .

But if you are running correct version, it looks weird. Please switch on debugging on miner and send me details about rejected hashes as PM. Thanks!

[zoidial]

How very right you are - that *is* a problem!  Glad the issue is that I'm an idiot and didn't update.

I just updated both m0mchil's miners on these two machines, and restarted (I had a version from the 21st of December) - and all hashes are accepted now.  Very odd how both machines started throwing invalid hashes at the exact same time still.  Glad the latest code fixes whatever issue that was, though.

Thanks for that suggestion!

[slush]

How very right you are - that *is* a problem!  Glad the issue is that I'm an idiot and didn't update.

Glad to read that! So all users which have invalid shares often, please check, if you are using latest version of miners!

Very odd how both machines started throwing invalid hashes at the exact same time still.

Yes, latest sources are without known problems. This issue was related with previous optimization in m0mchil miner, which is not compatible with pool because of more strict checks on pool side. So when m0mchil released version without 'changing ntime' feature and I turned on strict job checking on 5.January, things become broken for users on old version...

[BitLex]

all users which have invalid shares often, please check, if you are using latest version of miners!

as mentioned sometimes already,
running 5970-windows-systems, it's impossible (at least to me) to run the latest m0mchil versions on both cores,
it's not a problem (or less problematic, but it works) to run pretty old versions (<20101126 on both cores, or second core only) though.

last Diablo miner i tried worked fine on both cores, but lags my deskotp down too damn much to still use it efficiently.

[ElectricGoat]

I've been trying to create an account on slush's website for some days, but I never receive the activation emails (I tried various email addresses). Has any of you registered recently ?

[fabianhjr]

I still get like 20% of invalid or stale hashes. I am in the newest version and just double checked.

It is also a 5_8_70 not a 5970. xD

[slush]

I've been trying to create an account on slush's website for some days, but I never receive the activation emails (I tried various email addresses). Has any of you registered recently ?

Hi, I just checked registration and everything works fine. Please check your spam folders or use another email provider (not just another email address).

[slush]

I still get like 20% of invalid or stale hashes. I am in the newest version and just double checked.
It is also a 5_8_70 not a 5970. xD

Wow, it is really weird. Could you please try another miner?

[fabianhjr]

Well, I guess it is stale since someone submitted a similar hash before. It might be an issue with the way the hash generator is seeded. I will take a look at it and tell you later. It is not that much, it is more like 2 or 3 each 100.

[slush]

Well, I guess it is stale since someone submitted a similar hash before. It might be an issue with the way the hash generator is seeded. I will take a look at it and tell you later. It is not that much, it is more like 2 or 3 each 100.

No, that cannot be a reason. It's just about you are sending garbled data somehow.

[slush]

Today I blocked one user with 1000 registered workers and many hundreds of them active. Looks like first botnet on the pool, which is something I don't want to support. I'm thinking how to prevent this, because it consumes too much resources for nothing (from my first investigation, every worker had around 500-800 khash/s, but generated too much network traffic).

[BitLex]

I still get like 20% of invalid or stale hashes. I am in the newest version and just double checked.

It is also a 5_8_70 not a 5970. xD

i also got a lot of "invalid or stale" hashes last night from ~0:00-4:00 server-time (that's when i decided to leave the pool),
on both of my cards, HD5970 and HD5850, i doubt it's the miner (poclbm0104) though.

testing right now and it seems like everythings back to normal.

[fabianhjr]

Today I blocked one user with 1000 registered workers and many hundreds of them active. Looks like first botnet on the pool, which is something I don't want to support. I'm thinking how to prevent this, because it consumes too much resources for nothing (from my first investigation, every worker had around 500-800 khash/s, but generated too much network traffic).

well, the issue is that he can come back and this time register about 100 accounts and have 10 workers on each. Why must it "getwork" each 5 seconds? Wouldn't it be better to be each minute or five? Also, if one came expect more to come. This might start growing.

Also, what if it is just a collection of low end computers? I have a friend with like 50 old computers(The majority still works) which he freecycled.

I am doing better now. My hashrate is more constant with the new client. Enjoy your bigger 6% share now that I upgraded.

[teknohog]

Why must it "getwork" each 5 seconds? Wouldn't it be better to be each minute or five?

In my understanding, it is because we expect a new block every 10 minutes. If you have to crunch for 5 minutes, there is a good chance that someone else produces a fresh block first, and your block will be stale.

Nevertheless, it is appropriate to highlight these issues. I feel that huge mining pools are not good network-wise. In addition to the increased traffic, you have a single point of failure, in what used to be a distributed P2P effort.

I think pools are great for slower GPUs and CPUs, because otherwise the user might not be interested in mining at all. Thus their contribution can make the network stronger. However, when miners with strong GPUs switch from local mining to a pool, the effect is quite the contrary.

[slush]

i also got a lot of "invalid or stale" hashes last night from ~0:00-4:00 server-time (that's when i decided to leave the pool),
on both of my cards, HD5970 and HD5850, i doubt it's the miner (poclbm0104) though.

It would be nice if you can turn it on again and log data which throws invalid hashes. There must be some reason for that. I have to say I did not change anything in core in last many days (exactly until 5.January).

Afaik there can be those reasons for invalid hashes:
1. Old poclbm version
2. Overclocked card
3. Network connection troubles
4. Server troubles

I don't know about other reason why it should not work. From server graphs, I see one small drop in network transfers at 1:00 UTC (approximately for 5-10 minutes), which may indicate some network troubles, but I don't think it is significant. But I can confirm that some miners are trying to re-upload shares, which means they have troubles with their connection. But it is probably nothing what I can manage.

[slush]

Why must it "getwork" each 5 seconds?

I think technohog described it well. I'm thinking about push based protocol instead of current getwork, where server will send a job to client only on merkle hash change (so only few times per minute, in the worst case). But it will takes long time before it will be stable enough to switch pool to this.

Also, what if it is just a collection of low end computers? I have a friend with like 50 old computers(The majority still works) which he freecycled.

Nobody who owns 1000 old computers will let them compute hashes, because thanks to hash/W ratio it is completely worthless.

[ElectricGoat]

What about creating a small proxy server between your server and all the workers for a single user ?

I have three workers, and they all call getwork every 5 seconds. What if I had a proxy that they would each call, but that would only relay a single getwork call to your server ? It seems pretty easy to write, and could lower your server load by quite a bit.

[j16sdiz]

What about creating a small proxy server between your server and all the workers for a single user ?

I have three workers, and they all call getwork every 5 seconds. What if I had a proxy that they would each call, but that would only relay a single getwork call to your server ? It seems pretty easy to write, and could lower your server load by quite a bit.

a single getwork?
If you use a "normal" mining client, everbody would be working on the exactly the same hash -- this is a complete waste of time.

Of course we can change each miner start with different nonce.... or start with random nonce. This need changing a couple lines of code.