This can happen if for example there is a network policy in place preventing the communication due to the build pod being created in another node.
Check whether the delegate can reach the build pod (the one with container lite-engine) or if something might be blocking the communication (eg. a new network policy or a new node that was spun up which doesn’t encompass the previous network policy)?
This can also potentially happen in GKE where the property “Intranode Visibility” is improperly enabled. Attempt to disable it to see if things are working again.