A game that I liked (ChessBase 13)
[Event "Problemist of Ukraine"] [Site "?"] [Date "2018.??.??"] [Round "?"] [White "Rusz, Árpád"] [Black "(Study 149)"] [Result "1/2-1/2"] [SetUp "1"] [FEN "3R4/8/3P4/3K4/8/8/2rkr3/8 w - - 0 1"] [PlyCount "37"] [EventDate "2017.??.??"] 1. Ra8 $3 {After the rook was oddly placed in front of the passed pawn, now white seems to break another principle: "In rook endgames, it is usually better to choose the longer flank."} ({We have a symmetric initial position, why shouldn't be the only key move also symmetric? It seems so natural to push the passed pawn...} 1. d7 $2 Kd1 $3 (1... Rc3 $2 2. Re8 Rd3+ 3. Kc6 $11) (1... Re3 $2 2. Rc8 Rd3+ 3. Ke6 $11) 2. Rc8 (2. Re8 Red2+ 3. Ke6 Rc6+ 4. Ke7 Rc7 $19) 2... Rcd2+ 3. Kc6 Re6+ 4. Kc7 Re7 $19) ({Thematic try:} 1. Rh8 $2 Rc3 2. Rh4 Re8 3. Rd4+ Ke3 4. Rd1 Ke2 5. Rd4 Rd8 6. Ke6 Re3+ 7. Kf7 Rd7+ 8. Kf6 {Now both black's king and one of its rooks are on the longer flank where there is more space to manoeuvre.} Re8 {The quickest win.} 9. Rd5 Re4 10. Kf5 Rh4 $1 $19) 1... Re3 (1... Rc3 2. Ra2+ $11) 2. Ra4 $1 (2. d7 $2 Rd3+ 3. Ke6 Rc6+ 4. Ke7 Rc7 $19) 2... Rc8 (2... Rd3+ 3. Rd4 $11) 3. Rd4+ Kc3 4. Rd1 $1 {White has managed to transfer the rook behind its passed pawn!} (4. d7 $2 Rd8 $19) 4... Kc2 5. Rd4 Rd8 (5... Kc3 6. Rd1 {positional draw}) 6. Kc6 Rc3+ 7. Kb7 $1 (7. Kb6 $2 Rb8+ $19) 7... Rd7+ 8. Kb6 {Now both black's king and one of its rooks are on the shorter flank where there is less space to manoeuvre.} Rc8 9. Rd5 Rc4 10. Kb5 {Now black doesn't have a move like 10...Rh4! from the thematic try!} Rc3 11. Kb6 Ra3 $1 12. Rd4 $1 (12. Kc6 $2 Rda7 $1 13. d7 R3a6+ 14. Kc5 Ra5+ 15. Kd4 Rxd5+ 16. Kxd5 Rxd7+ $19) 12... Kc3 13. Rd1 Kc4 14. Kc6 Rda7 15. Rc1+ ({ Pushing the pawn would be fatal:} 15. d7 $2 R3a6#) 15... Kb4 16. Rb1+ Ka5 ( 16... Kc4 17. Rc1+ {perpetual check}) 17. Rb5+ (17. d7 $2 Rc3+ 18. Kd6 Rd3+ $19 ) 17... Ka6 (17... Ka4 18. Rd5 $11) 18. Rb6+ Ka5 19. Rb5+ {perpetual check} 1/2-1/2