A linker would need to be flexible, an inherently disordered; something like (Gly-Gly-Gly-Gly-Ser)n could have homologous recombination issues; GSAGSAAGSGEF would probably work better, but each protein has its own issues, so try and error...
Miguel Montoya GSAGSAAGSGEF is good choice for flexiable linker, you could also use different codon for the (GGGGS)n linker to avoid homologous recombinaiton. A linker doesn't always need to be flexiable, it could be rigid or cleavable as well, which depends on the protein structures and the experiment design. Here is a review for your reference. Article Fusion Protein Linkers: Property, Design and Functionality