search

B3-Coin / B3-CoinV2

B3-CoinV2
MIT License
46 stars 64 forks source link

Pose Score of fundamental node does not decrease as expected #52

Closed engzizo79 closed 5 years ago

engzizo79 commented 6 years ago

The nScanningErrorCount property should decrease when scanning is successful and increase if not. However, it seems that fns stay many hours at the same positive score. The debug log shows that pings arrive every 5 minutes as expected, but pose score stays the same. Issue has been noticed on remotely started fns on linux Ubuntu 16.04

engzizo79 commented 6 years ago

Also the new wallet version 3.1.2 always shows a pose score of 0 when the command 'fundamentalnode pose' is run. Edit: that is not correct, it started with 0 but it does change.

joshafest commented 6 years ago

PoS is type of state, yes it should decrease if scan is successful nice idea. Please make a proposal, and you will be added to credits

AltFreq07 commented 6 years ago

I have been monitoring my nodes over a few days. Every 4 hours or so the whole network should be scanned by the wallet either giving a positive nErrorType on a SCANNING_SUCCESS decreasing the nScanningErrorCount by 1 or for any other error increasing nScanningErrorCount by 1.

I have a few nodes that have a pose score or nScanningErrorCount > 0 which should be either decreasing by 1 or increasing by 1 at least every four hours according to a comment in fn-service.h These nodes have their pose score updated maybe once every 8-10 hours if lucky with majority of them not changing at all.

joshafest commented 6 years ago

Please, label enhancements

engzizo79 commented 6 years ago

I was wondering if this problem is related to the setting that wallets don't connect for 24 hours by default to wallets that were 'misbehaving', therefore it would actually take 24 hours + x for any decrease to happen?

engzizo79 commented 6 years ago

I am still trying to understand the code, from fn-service.cpp it seems that only 1% of the nodes are scanned per block, which would make it 100 blocks on average per scan, that is more than 10 hours and not 4. Also I'm not sure how it is controlled that the scanning node is not the one that has a problem with the connection when it broadcasts the problem? I think it would be important to increase the frequency of scanning the higher the pose score is to make recovery time reasonable. I don't know how rank is determined, and obviously only the top 1% in rank will be scanned each block

joshafest commented 5 years ago

I think the issue resolved, if not re-open it