Staking effectively disabled if no block for 60 minutes

[CaveSpectre11]

The problem code is in miner.cpp:

            if (!gArgs.GetBoolArg("-genoverride", false) && (GetAdjustedTime() - nTimeLastBlock > 60*60 || IsInitialBlockDownload() || !g_connman->GetNodeCount(CConnman::NumConnections::CONNECTIONS_ALL) || nHeight < Params().HeightPoSStart())) {
                MilliSleep(5000);
                continue;
            }

In commit 5b7a6a6 things started to go down this path; that results in the staking effectively being turned off if there's no block in 3600 seconds. The intent of genoverride here is to allow another mined block to unstick the chain.

genoverride is a developer tool; but the basic thing is that if there's a significant drop in staking power, and we hit the limit on mined blocks in a row; we should still wait for a stake, for however long it takes, without needing a gen-overridden mined block to come.

[Zannick]

Would there be any negative effect from just removing this time check? GetAdjustedTime() - nTimeLastBlock > 60*60

[CaveSpectre11]

Would there be any negative effect from just removing this time check? GetAdjustedTime() - nTimeLastBlock > 60*60

The part that just logically throws me for a loop is that what it -should- allow is: If there's been too many consecutive mining blocks so the next block -has- to be a staking block, but there's not been a block for 6 minutes, it re-allows mining for another block.

[Zannick]

The part that just logically throws me for a loop is that what it -should- allow is: If there's been too many consecutive mining blocks so the next block -has- to be a staking block, but there's not been a block for 6 minutes, it re-allows mining for another block.

That's going to be a consensus change, because we're not turning off mining but rejecting the pow block:

veil/src/validation.cpp

Lines 4667 to 4671 in 27d0b70

	if (!isNewBlock) {
	if (pindexPrev->nHeight >= Params().ConsecutivePoWHeight() && !CheckConsecutivePoW(block, pindexPrev)) {
	return state.DoS(100, false, REJECT_INVALID, "bad-pow", false, strprintf("too many consecutive pow blocks"));
	}
	}

The intent of genoverride here is to allow another mined block to unstick the chain.

I don't think that works, then, because this is only in staking code: the effect is only to allow the user of genoverride to create a stake after 3600 seconds. This would only be a change on the staking side, unless I missed a consensus check somewhere where we reject stake blocks after 3600 seconds (I don't think we do or we would have failed to revive testnet as much as we have).

Also, oops, 3600 seconds = 60 minutes, I don't know how I misread that before. That's a lot less worrisome than I initially thought. :)

[seanPhill]

Aha. 60 minutes gels with our experience of testnets getting stalled. I don't really mind so long as we remember to include some PoW when getting things running.