Cluster takes too long to apply cluster state

Hmm this sounds like a possible bug, we shouldn't be blocking cluster state updates on deletes no matter how large the indices are. Could you follow the troubleshooting guide to work out what extra information to collect to dig deeper?