[Fastacmd] How To Retrieve Sequence From Blast Db

Let's say I have multifasta with protein sequences having internal IDs (integer)

>1234
MGKL...*

I build blast db using:

formatdb -i infile.fa -pF -n someDB

But then, I'm unable to retrieve sequence from db using simple protein id:

fastacmd -d someDB -s 1234

How to define fasta header so I can retrieve sequences easily?
I have noticed formatdb assign internal identifiers (increment int) to my sequences, and orginal ID appears later:

>gnl|BL_ORD_ID|12 1234

Why is that?

I then defined headers as:

>gnl|dbname|1234

but with no effect. Do I have to define headers as >gi|1234 in order to be able to get sequence? Or is there any other way of retrieving sequences from blast db?