Testnet Issues: Header sync problem

[Barrett Davis]

We've been trying to write our dApp ontop of Dash, but testnet keeps giving us a head ache! Dashd keeps dying on our containers because of a "bad peer"?

2018-04-08 18:01:17 ERROR: detected bad peer for initial headers sync, disconnecting 5478

Every time we detect a bad peer, we infinite loop into a disaster.

Part of the log:

018-04-08 18:01:16 ThreadSocketHandler -- removing node: peer=5477 addr=116.62.210.72:19999 nRefCount=1 fNetworkNode=1 fInbound=0 fMasternode=0
2018-04-08 18:01:16 ProcessMessages: advertising address 52.91.249.51:19999
2018-04-08 18:01:16 receive version message: /Dash Core:0.12.2.3/: version 70208, blocks=100127, us=52.91.249.51:35156, peer=5487
2018-04-08 18:01:16 AdvertiseLocal: advertising address 52.91.249.51:19999
2018-04-08 18:01:17 ERROR: detected bad peer for initial headers sync, disconnecting 5478
2018-04-08 18:01:17 ProcessMessages(headers, 81570 bytes) FAILED peer=5478

 

[codablock]

This is a known issue at the moment and being worked on. Someone has plugged his ASICs into testnet and generated so many blocks in one row that it resulted in a fork. Not sure why exactly the fork happened, but it seems to be connected to superblocks and MN reward payments. Didn't have much time to investigate it, but UdjinM6 and flare are working on it atm.

 

[demo]

it seems to be connected to superblocks and MN reward payments....

Whenever problems occur in MN reward payments, I wonder about this.

 

[demo]

This is a known issue at the moment and being worked on. Someone has plugged his ASICs into testnet and generated so many blocks in one row that it resulted in a fork. .

What do you mean by "in one row" ?
How is it possible to create blocks in one row?
I assume you are using it as an expression and not literally.

 

[thephez]

If someone adds an ASIC to a system with low difficulty (like testnet), they vastly overpower the rest of the mining capacity. For instance, the testnet difficulty is frequently between 20-30. It then jumped to something over 700 - making it almost certain that the ASIC would mine every block.

 

[Barrett Davis]

Thanks for the great response guys. Is there any monitored channel that more real-time than a forum? Especially for testnet... if's its unstable I would love to check one spot to know or reach out on a github issue or something...

 

[thephez]

You might have better luck with the Discord (dashchat.org) #testnet-lab channel. I know this has been discussed over there too.