Rotate Backup Witnesses Into The Production Queue #14
Comments
|
Althought this seems a very noble gesture and it is, the way the graphene blockchain software is designed is that it is able to handle 1001 active witnesses. That is more than currently on offer. Backup witnesses can already become active witnesses if major PPy stakeholders are convinced about their knowledge and experience and therefore vote more witnesses in. Instead of changing the software its best to convince stakeholders to vote for more witnesses. Additionally simply by seeing whomever is missing blocks will show who is reliable and can backup witnesses be voted out as soon as this is noticed. Imho :) |
|
"Backup witnesses can already become active witnesses if major PPy stakeholders are convinced" How will the stakeholders know who to vote on if these servers never get tested? It seems that the current method is going to compound and enforce a massive barrier to entry resulting in small elite group of witnesses. The website hosts a video encouraging people to become a witness. New people are being encouraged to spend money in order to become a witness and the majority will not be able to earn anything back under the current system. With no trickle down blocks and only active witnesses being listed the system feels stacked against new people investing into the platform that were encouraged to do so as an opportunity to make money. This could result in resentment and negative sentiment towards the project being spread. |
|
I'm not familiar with the internals of Graphene or Peerplays and don't know if this is possible, |
|
You can setup a testnet for testing block signing! But rest assured if you have console output in the witness_node about receiving blocks with latency info that your witness is running. You can verify block nums with the GUI blocks. You can try and use cli_wallet on the live chain and if that works you know you are connected and synced.
Only things you could have made mistakes with then and that you cannot know for sure unless you are active witness (or triple check), are the following config.ini settings:
1. Did not use the correct witness-id in the config.ini but the account-id. Witness object ids start with 1.6.X so cross check that with `get_witness UrWitnessAccountName`
2. Did not use the correct pub,priv-keypair in the config.ini settings. Use `get_witness UrWitnessAccountName` and check the public signing key. Then use `get_private_key UrPubSigningKey` to get the priv-key. Now crosscheck in config.ini if you have that correctly set at `private-key=` and also the correct order in the array: `["pub","priv"]`
… On 11 Jun 2017, at 04:50, Alex Lobakov ***@***.***> wrote:
I'm not familiar with the internals of Graphene and Peerplays and don't know if this is possible,
but it would be nice to have a capability enabling backups/inactive witnesses to run in "shadow" mode. Essentially, they wouldn't be earning any PPY, nor would the blocks they produce be included in the production blockchain, however stakeholders would be able to see their statistics and make an informed decision regarding each inactive witness' performance and reliability, and subsequently vote those in.
Am I day-dreaming?
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
Bacchist's proposal is an acknowledgement of and solution to a real problem. It is superior to minimizing and justifying the problem. |
|
So what is the problem then? Inactive witnesses not verifying blocks to check whether their software is actually working?
So then what would happen if it would be like steem and in-active witnesses would once every couple of hours / day you are going to sit and wait until the block is signed and verified or failed and then start to debug potential problems and then restart your witness_node and wait another 24hrs to see if your fix was actually fixing the bug?
I think best would be to have more witnesses voted in as active and its supersimple. This is simply reached by reaching out to high stakers and asking them to vote for more witnesses. That way all active witnesses are signing blocks every one or 2 minutes. Once high stakers vote more witnesses in within an hour more are active and can "test" their setup.
It is a matter of getting yourself heard, promote yourself and reach out to high stakers to vote for more witnesses to become active.
i think the core software devs are currently more busy with writing the actual chain software for lauching betting / game dApps and so they should imho now the basic consensus graphene chain is in place.
Nonetheless you could write some code and test your hypothesis and submit a pull request by merging steem code and graphene code. Imho.
… On 11 Jun 2017, at 09:42, coloneldbugger ***@***.***> wrote:
Bacchist's proposal is an acknowledgement of and solution to a real problem. It is superior to minimizing and justifying the problem.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
Another to solution is: setup a proxy and gather many small & big ppy stakeholders to have their vote count through that proxy and then vote with that proxy on as many witnesses as possibke. Even extending it by writing some software for that specific proxy which removes votes on witnesses automatically by analysing witness performance.
Another solution: buy many ppy and become a high stakeholder :p
… On 11 Jun 2017, at 11:12, Roeland P ***@***.***> wrote:
So what is the problem then? Inactive witnesses not verifying blocks to check whether their software is actually working?
So then what would happen if it would be like steem and in-active witnesses would once every couple of hours / day you are going to sit and wait until the block is signed and verified or failed and then start to debug potential problems and then restart your witness_node and wait another 24hrs to see if your fix was actually fixing the bug?
I think best would be to have more witnesses voted in as active and its supersimple. This is simply reached by reaching out to high stakers and asking them to vote for more witnesses. That way all active witnesses are signing blocks every one or 2 minutes. Once high stakers vote more witnesses in within an hour more are active and can "test" their setup.
It is a matter of getting yourself heard, promote yourself and reach out to high stakers to vote for more witnesses to become active.
i think the core software devs are currently more busy with writing the actual chain software for lauching betting / game dApps and so they should imho now the basic consensus graphene chain is in place.
Nonetheless you could write some code and test your hypothesis and submit a pull request by merging steem code and graphene code. Imho.
> On 11 Jun 2017, at 09:42, coloneldbugger ***@***.***> wrote:
>
> Bacchist's proposal is an acknowledgement of and solution to a real problem. It is superior to minimizing and justifying the problem.
>
> —
> You are receiving this because you commented.
> Reply to this email directly, view it on GitHub, or mute the thread.
>
|
|
You didn't read this part of the problem? I will repaste: " It seems that the current method is going to compound and enforce a massive barrier to entry resulting in small elite group of witnesses. The website hosts a video encouraging people to become a witness. New people are being encouraged to spend money in order to become a witness and the majority will not be able to earn anything back under the current system. With no trickle down blocks and only active witnesses being listed the system feels stacked against new people investing into the platform that were encouraged to do so as an opportunity to make money. This could result in resentment and negative sentiment towards the project being spread." I get that you are a Graphene tycoon that can float from project to project with great support from your previous work but to expect others to be able to accomplish that feat is somewhat callous and disconnected from reality. Maybe the promotional materials should be updated to include the fact that if you've haven't already arranged votes from major stakeholders or aren't willing or capable to spend a fortune on stake yourself you will probably never make a penny back and shouldn't bother. With so much competition in this space (Augur, Gnosis, Wagerr) good witnesses and developers will flee and go elsewhere. Why would someone stick around to invest more and develop for a system stacked against them? |
|
Thx Colonel,
I feel that it is very early to say that this is a problem. If it would stay that way it definitely is. So yes a fix will be to list all witnesses in the GUI, I totally agree. Maybe currently the amount of active witnesses is just the result of one big stakeholder voting for (who knows) 2 or 3 witnesses because s/he saw this in the GUI this "funny option of voting" and hit 'add' once or twice or just for the list of displayed (active) witnesses.
So yes I feel your potential problem but to say it is a problem now already is very fast imho. PPY has only launched the GUI what, 2-3 days ago and now it is already expected that everything works just fine and what doesn't should already be adjusted by rigorously changing software and it's design... But then again I feel what you say if it is about this 'encouraging video' and then nothing happens with many witnesses, so maybe that should have been clearer in the video (should review it again myself).
*Nonetheless I still feel that a simple solution (unfortunately I can't be of more help as I personally only had a 60 PPY buy-in) is to identify some large stakeholders and discuss how they see this and get them convinced to vote as many witnesses as they like.* Because it is baked inside the chain software that within 1 maintenance interval it will just add more active witnesses if more big-stakers add more witnesses to their voting list. This is all when I say that there is no real problem: consensus should be reached through the voting. _So yes a problem could be that high stakers are unaware they can vote more people. That definitely should be fixed, so I think we can agree on this._
On a side note(s) but actually this should not be on github, nonetheless I feel a bit offended & humbled at the same time by the wording 'graphene tycoon', so i want to say this:
Can you only imagine how I felt when seeing that witness list on steem about 10 months ago, thinking woaaah, woaahhh what does this GV sign mean? Additionally I hope I still continue to add new works and not just sit and "float from project to project due to previous work". I think I add my fair share of new stuff for every chain I am part of, so it feels a bit of a personal 'bark' to me. You can see all my announcements what I do, check my github or my steem blogs. I was also suprised to see the current voting distribution for PeerPlays although I have been active for a long time running up to PeerPlays new launch and then was already fairly acquainted with the project long before the testnets had launched.
Lastly just a fiy: I had run for steem a full expenses paid witness node (dedicated) for 3 months just for software experiments and to help my fair share because i have server side experience, before I became an active witness and with for example bitshares I ran a paid witness node (dedicated) for half a year before becoming an active witness.
We can discuss this further via telegram:roelandp :)
… On 11 Jun 2017, at 21:07, coloneldbugger ***@***.***> wrote:
You didn't read this part of the problem? I will repaste:
" It seems that the current method is going to compound and enforce a massive barrier to entry resulting in small elite group of witnesses.
The website hosts a video encouraging people to become a witness. New people are being encouraged to spend money in order to become a witness and the majority will not be able to earn anything back under the current system. With no trickle down blocks and only active witnesses being listed the system feels stacked against new people investing into the platform that were encouraged to do so as an opportunity to make money. This could result in resentment and negative sentiment towards the project being spread."
I get that you are a Graphene tycoon that can float from project to project with great support from your previous work but to expect others to be able to accomplish that feat is somewhat callous and disconnected from reality.
Maybe the promotional materials should be updated to include the fact that if you've haven't already arranged votes from major stakeholders or aren't willing or capable to spend a fortune on stake yourself you will probably never make a penny back and shouldn't bother.
With so much competition in this space (Augur, Gnosis, Wagerr) good witnesses and developers will flee and go elsewhere. Why would someone stick around to invest more and develop for a system stacked against them?
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub <#14 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAPJALlzjlHNTGmTDZbsHBlhU3EBDI-9ks5sDDrlgaJpZM4N2Lbp>.
|
|
" I feel a bit offended & humbled at the same time by the wording 'graphene tycoon', so i want to say this: Can you only imagine how I felt when seeing that witness list on steem about 10 months ago, thinking woaaah, woaahhh what does this GV sign mean? " It is not my intention to offend, sorry if some of my emotion leaks through. I know you also put in a ton of work for Steemfest and earned your prominence in the community. I'm not saying you don't deserve a top spot. I do think though that the current setup isn't very fair or decentralized and could be construed as a bait and switch given the current marketing materials. I see this as a problem that needs to be fixed right away. I see this as incredibly important time which will cast how the system works out for years to come. Steem was a horrible launch and it is still dealing with the issues created by how a few people amassed staggering stakes for arguably not providing much. We should't want to recreate that imbalance yet seem to using a system design to make it even worse. Right now is like moments after the big bang where forces are being forged that will forever shape what comes after. We can see with every hour that passes a massive disparity growing between active and inactive witnesses. Even as the total VESTS voting for all witnesses was increasing across the board the active witness list shrank. The people receiving nothing have been campaigning and are earning votes yet the current system ignores the stake voting for them and sends all the rewards to very few. Let's be real, much of the list isn't based on merit at the moment. It is largely a Steem and Bitshares popularity contest. I'm not trying to be offensive and pick on anyone but I have to bring up the fact that one of the top active witnesses, who leads the pack in missed blocks, is a teenager that lost his previous Steem witness server because he was grounded by his parents for getting bad grades. I bring him up also because he is an example of how this purposed tweak would have shown his server was not configured properly prior to being thrust into the active list. Testnet is great for learning but it is not a true test that your production server is working and many would like to see at least one block signed in production before they can feel less anxiety. I think the exposure within the GUI wallet is one of the largest culprits right now but combined with this issue of no blocks leaving the elite witness list it is being exponentially made worse. I totally get wanting to focus on Dapps and making something useful out the chain but I'm of the opinion that this witnessing issue should come first since it get's worse every 3 seconds. PS: I will fully admit to sounding like a butthurt whiner but that's kind of my point. |
|
It would be nice to get an official stance on this issue from the PeerPlays developers. Are improvements to the witnessing system going to be worked on? |
|
One added benefit of this approach is the "wildcard" witness actually adds additional manipulation resistance. If the top witnesses are all colluding, it's the wildcard that is used in order to submit the unapproval votes necessary to remove colluding witnesses (assuming a terrible event in which the "so called" elite are all colluding). Having said that, I'm sure the PBSA team has a list of priorities right now and will choose the right time to potentially include this feature. |
Similar to Steem, it would be beneficial for witnesses outside of the active group to be cycled through regularly. While we may not yet need the excess capacity, in the future these witnesses/nodes may be called upon to produce. Voters and node operators should be able to identify which backups are configured properly and reliably.
As applications are built out and witness feeds become a critical point of failure, it will become even more important to know which backups are ready to assume active production with reliable feeds as well.
The text was updated successfully, but these errors were encountered: