there's probably a mismatch between your pod termination grace period and the

--grpc-shutdown-grace-period

flag on spicedb