Author Topic: Forging Service - Anyone Interested?  (Read 666 times)

warmach on: November 12, 2016, 03:00:30 PM

  • Senior Fimker
  • ***
  • Posts: 234
    • View Profile
As we see our network stability becoming an issue, I'm trying to think of ways to help.  As we've determined, the instability is likely because we don't have a large number of forgers. 

I have an idea to help with that.  Before I write the code to make it happen, I wanted to ask the community if they wanted it and/or if it would be helpful.  Please respond with your feedback.

My idea is to do away with a forging pool and become forging service instead.  Each "client" would be given their own forging address.  The password to this address would be held by my service and the client.  The client would then lease their balance to their own forging address.  My server would then forge using the forging address.  After a set period of time, probably 24 hours, any forged funds would be transferred back to the client's original address. 

With this method, my pool would add another 15-20 forgers to network.  With both parties having the password to the forging account, trust required in the service is reduced as the client has direct access to their forged funds. 

Warmach's Forging Pool
http://fim.warmach.info/pool/
May the forge be with you...

eRoh #1 on: November 13, 2016, 10:32:08 AM

  • Grand Fimker
  • ****
  • Posts: 324
    • View Profile
I'm not sure, if that would help. If I've understood correctly, the number of servers (and the distribution of balances between the servers) is more relevant than the number of forging accounts. If you have multiple accounts forging on the same server, wouldn't they all end up in the same fork*? At least the opposite is true: I have one account forging on 2-3 servers at a time, and the servers end up in forks independently of each other. And only the "correct" one is shown in lompsa.com, of course.

What might work better, could be that warmach and/or any other "too" large server split in two, dividing the balance between two servers. For a pool, it would require moving half of the forgers (balances) to another server. For starters, warmach could move his own balance forging to another (account and) server. Also the forging support from association could run on multiple servers (if not already running). What do you think?

*) I'm using the term fork here in a wider sense: any chain of blocks, regardless of the "correctness" of that particular chain in consensus of the network.
FIM-34B3-PWY5-TMHC-9JYW8

warmach #2 on: November 14, 2016, 01:00:45 AM

  • Senior Fimker
  • ***
  • Posts: 234
    • View Profile
I'm not sure, if that would help. If I've understood correctly, the number of servers (and the distribution of balances between the servers) is more relevant than the number of forging accounts. If you have multiple accounts forging on the same server, wouldn't they all end up in the same fork*? At least the opposite is true: I have one account forging on 2-3 servers at a time, and the servers end up in forks independently of each other. And only the "correct" one is shown in lompsa.com, of course.

The more I think about it, the more I think you are right.  One server can forge with multiple addresses, but each server has its own local blockchain.  What we really need is more individual blockchain (a.k.a. servers) so that there is a greater consensus on the chain and less chance of forks.

What might work better, could be that warmach and/or any other "too" large server split in two, dividing the balance between two servers. For a pool, it would require moving half of the forgers (balances) to another server. For starters, warmach could move his own balance forging to another (account and) server. Also the forging support from association could run on multiple servers (if not already running). What do you think?

Well, I have a primary server that forges for the pool, but I also have backup servers.  Each backup server also forges in the event that primary server goes down or hits a fork.  If I have 3 servers running, I'm not sure it matters if I split the forging accounts among them or just use the pool account on all three.  In the end, I think it is how many server nodes are on the network, not how the forging accounts are spread out among them.

One question I do have, If I have 3 servers running behind one public IP, do we get the stability effect of 3 nodes?  Or does it just act as one?
Warmach's Forging Pool
http://fim.warmach.info/pool/
May the forge be with you...

eRoh #3 on: November 14, 2016, 06:44:06 AM

  • Grand Fimker
  • ****
  • Posts: 324
    • View Profile
The more I think about it, the more I think you are right.  One server can forge with multiple addresses, but each server has its own local blockchain.  What we really need is more individual blockchain (a.k.a. servers) so that there is a greater consensus on the chain and less chance of forks.

Well, I have a primary server that forges for the pool, but I also have backup servers.  Each backup server also forges in the event that primary server goes down or hits a fork.  If I have 3 servers running, I'm not sure it matters if I split the forging accounts among them or just use the pool account on all three.  In the end, I think it is how many server nodes are on the network, not how the forging accounts are spread out among them.
I think that besides the number of servers/blockchains, the "equality" of them is also important. If you split an "overpowered" balance to two servers, that should even out the "voting". But I'm on thin ice here, maybe Dirk or somebody can clarify...

Quote
One question I do have, If I have 3 servers running behind one public IP, do we get the stability effect of 3 nodes?  Or does it just act as one?
I've sometimes suspected that having multiple servers behind one NAT can cause problems, but I'm not sure if it was because of that or something else.
FIM-34B3-PWY5-TMHC-9JYW8