how to make synchronization inside a nested loop?

#pragma omp parallel num_threads(min_thread) shared(count){

#pragma omp for

        for (int i = 0; i < v_edgeWeight.size(); i++)        {                    // light cpu work                          for (long j=0; j<currentNedges; j++)            {                             // light cpu work                              if ( valid )                {                                  //wait all threads                                   // execute a line once and release all threads                                     }                            }                                                       }        }

i made a function with #pragma omp barrierand #pragma omp single

but the problem is that some threads get stuck at the barrier when some threads exits the loop without reintering the inner loop(which has to be sequential)