Pages:
Author

Topic: [NR 1] Triplemining.com <> BIG jackpot every week <> - page 18. (Read 113533 times)

donator
Activity: 2058
Merit: 1007
Poor impulse control.
Actually, i meant this ->

Quote
Another incorrect implementation is to pay all shares in a window given in units of time (sometimes called PPLNM or PPLNH). In addition to the problems above, it has a problem common to all reward systems that use time as a factor, which is that it is more profitable to mine when the current hashrate is higher than the average.

Please hop the pool, what you are saying is that the pool is more profitable if the hashrate rises.  When you join, the hashrate rises, so basicly you're saying that everybody should join the pool.

There is a difference between theoretical exploits and something we should worry about...  I consider all this FUD, while it is nice to know the underlying mathematics, and while there are real pitfalls to evade, some issues are irrelevant imho....

OK then. But keep in mind this is not theoretical and you can model it and the likely outcomes for yourself (this is me not trying to hard to convince you now that prop pools are on their way out and I'll have to hop the less hoppable pools)
sr. member
Activity: 263
Merit: 250
Pool operator of Triplemining.com
Actually, i meant this ->

Quote
Another incorrect implementation is to pay all shares in a window given in units of time (sometimes called PPLNM or PPLNH). In addition to the problems above, it has a problem common to all reward systems that use time as a factor, which is that it is more profitable to mine when the current hashrate is higher than the average.

Please hop the pool, what you are saying is that the pool is more profitable if the hashrate rises.  When you join, the hashrate rises, so basicly you're saying that everybody should join the pool.

There is a difference between theoretical exploits and something we should worry about...  I consider all this FUD, while it is nice to know the underlying mathematics, and while there are real pitfalls to evade, some issues are irrelevant imho....
donator
Activity: 2058
Merit: 1007
Poor impulse control.
Not quite. You do know that PPLNH is hoppable due to the time based nature of the PPLNH score?

I  have not found any evidence, can you provide a link to a technical explanation?


It's an explanation, but not a technical proof:

[url]https://bitcointalk.org/index.php?topic=39832.0/url]

Oh that case.  Right.  Yes it is correct that all PPLNS pools are hopable by hopping correctly when the difficulty changes.  However, the difficulty changes are very low, and only once every 2 weeks.  So yeah, if you really want you can do it but the effort is not worth the profits.

Actually, i meant this ->

Quote
Another incorrect implementation is to pay all shares in a window given in units of time (sometimes called PPLNM or PPLNH). In addition to the problems above, it has a problem common to all reward systems that use time as a factor, which is that it is more profitable to mine when the current hashrate is higher than the average.
sr. member
Activity: 263
Merit: 250
Pool operator of Triplemining.com
Not quite. You do know that PPLNH is hoppable due to the time based nature of the PPLNH score?

I  have not found any evidence, can you provide a link to a technical explanation?


It's an explanation, but not a technical proof:

[url]https://bitcointalk.org/index.php?topic=39832.0/url]

Oh that case.  Right.  Yes it is correct that all PPLNS pools are hopable by hopping correctly when the difficulty changes.  However, the difficulty changes are very low, and only once every 2 weeks.  So yeah, if you really want you can do it but the effort is not worth the profits.
donator
Activity: 2058
Merit: 1007
Poor impulse control.
Not quite. You do know that PPLNH is hoppable due to the time based nature of the PPLNH score?

I  have not found any evidence, can you provide a link to a technical explanation?


It's an explanation, but not a technical proof:

https://bitcointalksearch.org/topic/pplns-39832
sr. member
Activity: 263
Merit: 250
Pool operator of Triplemining.com
Not quite. You do know that PPLNH is hoppable due to the time based nature of the PPLNH score?

I  have not found any evidence, can you provide a link to a technical explanation?
donator
Activity: 2058
Merit: 1007
Poor impulse control.
Triplemining is switching to PPLNS, N=24h

As we've discussed before, proportional payouts are not the way to go, and we've discussed PPLNS as a better alternative for our payout system.

We've been working on the idea, and we've decided to switch the pool to PPLNS, where N = 24h effective from the next found block.

This is not PPLNS as other pools implement it, by using the last 24h, people who can only mine at night for example; can have full profit of found blocks.

The new backend is tested and ready, the website will be changed the days following the next found block.

This is good news for the loyal miners of our pool, as the payout will only increase for those committed to mine with us.

Not quite. You do know that PPLNH is hoppable due to the time based nature of the PPLNH score?

You might do better with DGS since you can adjust your own risk, your miners variance and is provably unhoppable. Also you wont have to keep track of shares, just one share per miner.
full member
Activity: 190
Merit: 100
Par Pari Refertur
Triplemining is switching to PPLNS, N=24h

As we've discussed before, proportional payouts are not the way to go, and we've discussed PPLNS as a better alternative for our payout system.

We've been working on the idea, and we've decided to switch the pool to PPLNS, where N = 24h effective from the next found block.

This is not PPLNS as other pools implement it, by using the last 24h, people who can only mine at night for example; can have full profit of found blocks.

The new backend is tested and ready, the website will be changed the days following the next found block.

This is good news for the loyal miners of our pool, as the payout will only increase for those committed to mine with us.




Nice, hope my income on the pool increases. Smiley
@kinlo - My laptop is coming tomorrow from service. Wink
sr. member
Activity: 263
Merit: 250
Pool operator of Triplemining.com
Triplemining is switching to PPLNS, N=24h

As we've discussed before, proportional payouts are not the way to go, and we've discussed PPLNS as a better alternative for our payout system.

We've been working on the idea, and we've decided to switch the pool to PPLNS, where N = 24h effective from the next found block.

This is not PPLNS as other pools implement it, by using the last 24h, people who can only mine at night for example; can have full profit of found blocks.

The new backend is tested and ready, the website will be changed the days following the next found block.

This is good news for the loyal miners of our pool, as the payout will only increase for those committed to mine with us.


sr. member
Activity: 462
Merit: 250
I heart thebaron
I will definately do that with the next block.

Have you had any problems lately?  You haven't responded yet...
No sorry. I actually began merged mining on another pool, but what I can tell you is oddly enough I ran into the exact SAME problem on the pool I am on now as I experienced on Triplemining (lots of stales, only helped using a short work scan time, making me very inefficient with lots of dropped/unsubmitted work) and it turned out to be a routing problem.....as in, the location of the server in question that I was connecting to. The Pool OP then provided me with an alternate server in a different geographical location and everything was solved, although others seemed to connect without consequence. ISP conflicts perhaps.

Just thought I would share, as it seemed to answer my question.....on both fronts.

Cheers,
Allan
sr. member
Activity: 263
Merit: 250
Pool operator of Triplemining.com
I will definately do that with the next block.

Have you had any problems lately?  You haven't responded yet...
sr. member
Activity: 462
Merit: 250
I heart thebaron
Best thing to do for these type of problems is to find me online on our irc channel #triplemining on freenode.  I do not know what is going wrong here, but if you want to help, we can do some testing to see what is exactly going on by monitoring what you are sending to the server and what the server is replying, so we can find the root cause of the problems.

I will definately do that with the next block.

Another thing I forgot to mention that I just remembered is that at the START of a block, things seem to go quite well, but as the block progresses things seem to go downhill for me, as I described above.

Has anyone else noticed degraded performance as the block goes on longer ? It's like if the block doesn't get solved quickly, performance degrades over time. I don't think this is an issolated incident, as this might explain why the pool either has extremely SHORT blocks or extremely LONG blocks and nothing in between as it were.....I could be wrong though. See you next block.
Allan
sr. member
Activity: 263
Merit: 250
Pool operator of Triplemining.com
Hey Kinlo, I use CGMiner and I get a retarded amount of stales from Triplemining for some reason. There are a couple other pools that are similar, but I can also achive under 1% from a few others.

At 1 point, I actually had to set my work-scan-time to 10 seconds in order to avoid submitting stales constantly, so if a hash couldn't be achived in 10 seconds, it would move onto the next attempt.

Could you shed some light on why this might be and why my temporary fix worked ?

Could this have something to do with the fact that CGMiner is multi-threaded and has the ability to use a single account for an entire motherboard (if setup that way....4 cards, 1 account?)...just thought I would add that.

Best thing to do for these type of problems is to find me online on our irc channel #triplemining on freenode.  I do not know what is going wrong here, but if you want to help, we can do some testing to see what is exactly going on by monitoring what you are sending to the server and what the server is replying, so we can find the root cause of the problems.
jr. member
Activity: 59
Merit: 1
Honk, honk!
Thank you for the excellent explanation. I did think that the stats reporting was a temporary issue and knew all would be fine, I only raised the query after some 48 hours of noticing it to make sure it was on the record. No-one can argue with 100% uptime on fallback  Grin
sr. member
Activity: 462
Merit: 250
I heart thebaron
Hey Kinlo, I use CGMiner and I get a retarded amount of stales from Triplemining for some reason. There are a couple other pools that are similar, but I can also achive under 1% from a few others.

At 1 point, I actually had to set my work-scan-time to 10 seconds in order to avoid submitting stales constantly, so if a hash couldn't be achived in 10 seconds, it would move onto the next attempt.

Could you shed some light on why this might be and why my temporary fix worked ?

Could this have something to do with the fact that CGMiner is multi-threaded and has the ability to use a single account for an entire motherboard (if setup that way....4 cards, 1 account?)...just thought I would add that.
sr. member
Activity: 263
Merit: 250
Pool operator of Triplemining.com
Hiya. Sorry to be a pest, but it appears that my last 12,000 or so shares have not been appearing in my account or workers pages on the website for eu2.triplemining.com - the connection is fine, and only 3 rejected, just wanted to check if you are on top of it. They will update once a new block is found?

Let me explain how triplemining works, before answering your question.

We have 2 nodes, eu1 and eu2 (eu is an alias for eu1).  We've designed the system so that in any possible case, the pool cannot be down.  How do we achieve this:  each node has it's own database, with a copy of the user table, and it stores the found shares locally.    In a separated process, these shares are replicated to our main node that runs the main site.  If for any reason, servers go down or anything else goes terribly wrong, eu1 and eu2 can continue on their own as they have no dependencies on anything else.  So, up till now, we've can proudly say that our pool has had a 100% uptime since we introduced eu2 for those miners that use eu1 and eu2 as failover for each other.  eu1 and eu2 are on completely different networks in different datacenters, several hundreds of kilometers seperated.  This will make it very unlikely that our pool will ever go completely down, as long as miners use both in a failover method.

This does mean that if something goes wrong, the first thing to break is the synchronization between the servers.  This is not a real problem, as the code is aware of such possibility, and if a block is found, the payout code will wait until both clusters have been synchronized fully before doing any payouts or calculations.  The downside is that if something breaks, the statistics will be incomplete, waiting for the shares found on either cluster to be fully synced.

We have had some issues with eu2 a few days ago, outside our control, which have forced us to temporary disable our monitoring on the eu2 cluster.  We've forgot to re-enable monitoring after these issues were resolved, causing us not to notice that eu2 was not syncing to the main servers.  This caused your shares not to appear on your pages.   As the code would not do any calculations when a block is found, this is only a "cosmetic" problem, we can guarantee that only when all shares are reported, the calculations and payouts would have been done.

I've restarted the syncing, so all statistics are correct again.

For those mining with triplemining, as a reminder: please configure your miner to use eu1 as main pool, and eu2 as backup pool if it is unavailable, or the other way around, eu2 as main pool and eu1 as backup. This guaranties that your miner will always be able to work on our pool without any problems, even if our servers go down, as it is highly unlikely that both datacenters become unreachable at the same time.
jr. member
Activity: 59
Merit: 1
Honk, honk!
Hiya. Sorry to be a pest, but it appears that my last 12,000 or so shares have not been appearing in my account or workers pages on the website for eu2.triplemining.com - the connection is fine, and only 3 rejected, just wanted to check if you are on top of it. They will update once a new block is found?
jr. member
Activity: 59
Merit: 1
Honk, honk!
Impressive!

edit

6 blocks in 1 day = whaaaaaaaaaaaaa?!!11111111
sr. member
Activity: 263
Merit: 250
Pool operator of Triplemining.com
Wheeee, our shortest block ever!!   3 blocks in just about one hour!

Go Go Go Go Go!!
sr. member
Activity: 263
Merit: 250
Pool operator of Triplemining.com
Im connected to eu2 with two workers, but not showing as updating in the workers section on the website (says 22 hours ago was last submit) - I assume this is because of the recent breakages and actually my shares are not being lost?
This was indeed related to the recent problems.  The mining servers keep their shares stored locally, and report them to a central server for statistics, which are used on the site.  When something goes wrong, they are not stored centrally, and the mining servers need to push them to the central server afterwards.  While during this process nothing is lost, you might see incorrect numbers on the site while the mining servers did not have the time to push everything into the central server.
Pages:
Jump to: