[net] Skip stale tip checks in regtest #12484

Starting a regtest node and leaving it for a bit would start spamming

2018-02-19 20:05:25 Potential stale tip detected, will try using extra outbound peer (last tip update: 32809 seconds ago)

which seems unnecessary. I check if it's regtest by seeing if Consensus::Params.fPowNoRetargeting is true, which is perhaps not ideal, but works.

Tend to NACK. That makes it impossible to test it on regtest, which is meant for testing

On Feb 19, 2018 20:11, "kallewoof" notifications@github.com wrote:

Starting a regtest node and leaving it for a bit would start spamming

2018-02-19 20:05:25 Potential stale tip detected, will try using extra outbound peer (last tip update: 32809 seconds ago)

which seems unnecessary. I check if it's regtest by seeing if Consensus::Params.fPowNoRetargeting is true, which is perhaps not ideal, but works.

You can view, comment on, or merge this pull request online at:

#12484 Commit Summary

• [net] Skip stale tip checks in regtest

File Changes

• M src/net_processing.cpp https://github.com/bitcoin/bitcoin/pull/12484/files#diff-0 (4)

Patch Links:

• #12484.patch
• #12484.diff

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/bitcoin/bitcoin/pull/12484, or mute the thread https://github.com/notifications/unsubscribe-auth/AGGmvx79K6bsaI84rMZMy-jBq-dnDvAmks5tWhutgaJpZM4SLSre .

@MarcoFalke All tests pass, though..

Edit: I see what you're saying. The spam is pretty obnoxious though.

Edit 2: Is this even testable in regtest? It seems kind of tied to a chain with PoW.

The stale tip check seems based on time, not PoW. So it could be potentially tested using mocktime on regtest.

I don't think log spam while testing is a very urgent issue, but if so, my preferred solution would be to add some kind of congestion control to the message - e.g. have it logged at most once per block, or in a certain time window, instead of disabling the functionality on a per-chain basis.

@laanwj @MarcoFalke I'm tempted to add a flag for whether you should check for stale tips or not, and make it default true for all but regtest. Tests could then explicitly enable it when testing the feature -- which no tests have done so far, mind you.

Downside is yet another flag to the startup options for the node / the .conf file.

which no tests have done so far, mind you.

Feel free to add one :) I'd much prefer that to any amount of hacking around this feature.

@laanwj That's the thing, though. Maybe I'm lacking in imagination but I can't think of a meaningful test for this. I'll give it some thought though.

Like @MarcoFalke , I'm NACKish on just disabling this on regtest, since we want regtest behaviour to be as close to mainnet behaviour as possible. Adding a flag seems reasonable enough, although I'd prefer the defaults to be the same for main/test/regtest.

Closing for now, until a better solution has been found.