Consider adding an ignore_tip or similar config option for forked chains

[oxarbitrage]

When forking the zcash testnet chain there is a point in the process where you need to comment out code that checks the chain is close to the tip such as:

https://github.com/ZcashFoundation/zebra/blob/v1.4.0/zebra-rpc/src/methods/get_block_template_rpcs/get_block_template.rs#L179-L199

The Qedit team reported this but i had to do the same when creating a testnet fork. I was wondering if it will worth to add a debug ignore_tip config option that could ignore that branch (and potentially others) when activated.

[teor2345]

I think that's a good idea. And it should say that Zebra is near the tip regardless of the internal service that's asking.

It could work the same way as the mempool option, which is debug_enable_at_height:

zebra/zebrad/src/components/mempool.rs

Line 279 in 382fb37

debug_enable_at_height: config.debug_enable_at_height.map(Height),

So for example in the QEDIT early testnet fork, they would set it to 20,000, or whatever the block before their fork height is.

[teor2345]

As a workaround for now, QEDIT could sync the full testnet chain, and then their fork would be near the tip.

[arya2]

They could reset the system clock, get the future blocks and submit them without peers, or start Zebra with peers to avoid commenting out that line, or fork the chain after the max checkpoint height to avoid deleting any checkpoints.

[oxarbitrage]

As a workaround for now, QEDIT could sync the full testnet chain, and then their fork would be near the tip.

It will be out of sync at some point anyways specially if you are experimenting, the workaround for now has been commenting out the code.

[arya2]

It will be out of sync at some point anyways especially if you are experimenting

Yeah it would be difficult to use debug_stop_at_height near the tip since the closest the cached state could be is depth 100 and the lowest acceptable depth for the getblocktemplate RPC is 101.

They could reset the system clock, get the future blocks and submit them without peers, or start Zebra with peers, but this workaround seems simpler. (Though they could still get around deleting the checkpoints by syncing past the max checkpoint)

[mpguerra]

Any outcome yet on this discussion? Should we create an issue from it?