BitShares Forum
Main => General Discussion => Topic started by: emski on July 26, 2014, 01:49:08 pm
-
1. I've noticed significant amount of forks lately. Any particular reason ?
2. I've noticed that one of the delegates misses blocks (3 so far) while the client is online.
Here is the last missed block:
60948
211d78002ecf8ad2b4b49c9a43b157e2064646ae angel.bitdelegate 0 166 2014-07-26T13:20:10 0 YES NO
a3c1f207ced3bc2b57869ed2661077ffea4207b4 mr.agsexplorer 0 166 2014-07-26T13:20:00 12 YES YES
List_errors mentions some exceptions but they are related to blocks 7531 and 47087 and 14872.
There are few more delegates working properly on the same client that are not missing blocks at all. Chances for this to happen randomly are 1/216. Is that particular delegate just unlucky or there might be something else?
3. Info reports "blockchain_average_delegate_participation": "100.00 %" , How is this calculated ?
-
I've noticed significant amount of forks lately. Any particular reason ?
I've noticed that one of the delegates misses blocks (3 so far) while the client is online.
Here is the last missed block:
60948
211d78002ecf8ad2b4b49c9a43b157e2064646ae angel.bitdelegate 0 166 2014-07-26T13:20:10 0 YES NO
a3c1f207ced3bc2b57869ed2661077ffea4207b4 mr.agsexplorer 0 166 2014-07-26T13:20:00 12 YES YES
List_errors mentions some exceptions but they are related to blocks 7531 and 47087 and 14872.
There are few more delegates working properly on the same client that are not missing blocks at all. Chances for this to happen randomly are 1/216. Is that particular delegate just unlucky or there might be something else?
Info reports "blockchain_average_delegate_participation": "100.00 %" , How is this calculated ?
How do you identify forks?
-
Just blockchain_list_forks not sure how are they collected though.
-
I noticed that someone's has registered my delegates name minus 1 letter. I am sure this is in attempt to collect funds based on someone mistyping or misspelling my name. My delegates name is skyscraperfarms and someone registered skyscraperfarm.
Be careful and make sure you have the correct spelling of the person you are intending to send to.
-
I was causing the forks as my backupsystem got activated accidentally .. sorry for that .. will implement some security in my scripting soon
-
I was causing the forks as my backupsystem got activated accidentally .. sorry for that .. will implement some security in my scripting soon
I hope there is an automated system in the client in the future. I have got two machines but have block production enabled only on one.
-
BTW when the blockchain_list_forks command mentions a block number
it is not the right one... the right one is + 1 block.
when it mentions for example block 51806 it is in reality 51807!
-
BTW when the blockchain_list_forks command mentions a block number
it is not the right one... the right one is + 1 block.
when it mentions for example block 51806 it is in reality 51807!
The fork is the block for which there are 2 next blocks.
-
I've seen strange pattern in the forks lately:
85154
e86dd2ad4549980139f65792045a66eb5f5f5119 c.delegate.xeroc 0 166 2014-07-29T11:11:50 0 YES NO
740472051be51c9d7824c9e025a5d6455be23b15 init97 0 166 2014-07-29T11:11:40 81 YES YES
85180
f540b93a91cde65690b2174e0d82ac7b554e7bc2 emski 0 166 2014-07-29T11:17:00 0 YES NO
fd900b79e1ba455f7b9677ece3ffc4ea7951b0e0 init86 0 166 2014-07-29T11:16:40 21 YES YES
85285
9832840513ca59d3fd4e463b94bb56807438867f boombastic 0 166 2014-07-29T11:36:10 0 YES NO
b27357e740273c7d4797df31a5754a72a438cae1 init86 0 166 2014-07-29T11:36:00 70 YES YES
85337
b6f6e73dc7e9a3c4ef0b77d6d092bf15dac8b5b9 delegate-watchman 0 166 2014-07-29T11:45:50 0 YES NO
ca6ed48bb428a74ec0d8d70d69a8dbe7b5300a5c init88 0 166 2014-07-29T11:45:30 21 YES YES
There are delegates with high latency and with earlier slot that are included in the chain.
However other delegates that produced and with low latency but at a later timeslot are not included in the chain.
How could this happen ?
-
Couple questions:
1. How do you check delegate latency
2. What does the ntp_error mean? What is good and what is bad?
-
Couple questions:
1. How do you check delegate latency
2. What does the ntp_error mean? What is good and what is bad?
The data you see in the above post is provided by blockchain_list_forks.
ntp_error means the difference in your server clock than the one calculated by client. This is going to be obsolete as I've seen bytemaster posting about removing it. You should install ntpd.
-
Another question:
97938
40cd6d227b22d64e4462c69bfc1e3a51bd1312af lotto.bitdelegate 0 166 2014-07-30T23:28:10 0 YES NO
292ee282ffb1b366442b9114ffc9b32933f00048 btsx.chinesecommunity 0 166 2014-07-30T23:28:20 -10 YES YES
98209
3824eb804261540b5a538fdf91f7af4655fc425b lotto-delegate 0 166 2014-07-31T00:15:30 0 YES NO
0810b0cb42cccb84895470b6a7dd62baeebc284b btsx.chinesecommunity 0 166 2014-07-31T00:15:40 -10 YES YES
98328
a424ad53d7a0feeb0a594a89d08cf314a902f90c btsx.chinesecommunity 0 166 2014-07-31T00:36:20 -10 YES YES
831953205b3982f0545d6330e059370a62286b52 a.delegate.charity 0 166 2014-07-31T00:36:10 0 N/A NO
98388
e42abc5a02e7ff5f1ae0e28196b70b8fc8907a67 btsx.chinesecommunity 0 166 2014-07-31T00:46:30 -10 YES YES
7a3f2e7fa820e555a29676d2e761b4ce17013e90 liondani 0 166 2014-07-31T00:46:20 0 N/A NO
What exactly happens in these forks?
-
I also got a seemingly strange one:
97525
3e9d0e1a61194b490a2b874f1133c85088cdda6f wackou-delegate 0 166 2014-07-30T22:15:50 0 YES NO
73c52ca0e0838712c8662d82c5cc301bad0d8071 init72 0 166 2014-07-30T22:16:00 0 YES YES
seems like I was on time and everything (as is shown by latency=0, I also had a lot of connections at that time and this was an isolated event, subsequent blocks after this one were produced correctly), is it just that somehow init72 "decided" to not include my block? Could it be that my block didn't propagate in less than 10 seconds to that node? In that case, it would probably be a good idea to have a "backbone" of broadcasting nodes, delegates and/or proxy nodes for delegates not wanting to broadcast their IP. I remember seeing this idea somewhere, maybe this has to be done in order for delegate to not randomly miss block (when they in fact correctly produced it)
-
AFAIR BM is working on a private network for active delegates .. I guess its just a matter of time
-
Great! Although that should probably just be a temporary solution until a better solution is found out, because:
- people will accuse us of forming a cabal
- this somehow centralizes the delegates, as 1 bad delegate could harvest the IPs of all the other ones easily, which would make it easy to DDOS a good part of them
I'm thinking having some types of nodes which are low-latency and non-peer-IP-forwarding to which delegates could connect which could act as a sort of backbone for instant and anonymous block broadcasting could be a good solution. I'm still not completely convinced, but I'll post something more detailed if I can get it clearer in my head.