Currently we don't support warp sync in parachains. We will need to write some Warp sync implementation for Parachains. The idea is that we first warp sync the relay chain and then get the latest header of the Parachain. After we got this header we can use StateSync on the Parachain side to get the state.

Depending on how the relay chain node is integrated, we can do the relay chain warp sync differently:

• In process relay chain node: Do a normal warp sync and then wait for the state sync to finish to extract the Parachain header. This process could be made faster by get the latest Parachain header through a storage proof from a relay chain node that already has the state, but this is more an optimization step.
• Out of process relay chain node: We would probably wait until the node is finished with syncing and then extract the latest Parachain header. While waiting we should probably print something like "Waiting for connected relay chain node to finish syncing".