I am assuming your running a cluster and you are configuring out-if-band management (fxp0)?
If so, there are number of reasons this will happen, for example: https://kb.juniper.net/InfoCenter/index?page=content&id=KB17161&actp=METADATA
If you search and browse the juniper articles your will probably find your problem, but a simple work around will be to login to either node from the other:
{primary:node0}
lab@host> request routing-engine login node 1