glusterd: volume start doesn't fail with properly if the port range is exhausted

[nik-redhat]

Description of problem:
gluster v start <volname> doesn't fail correctly if the port range is exhausted (or after limiting the value of max-port in the glusterd.vol file)
It should have failed with a correct error message i.e, "All the ports in the range are exhausted" in the log file.
glusterd hangs in the above scenario and needs to be restarted with the volfile updated to allow all the bricks to be started.

RCA
After the latest update to use random ports in pmap, we are executing the below code to check for free ports:

int
pmap_port_alloc(xlator_t *this)
{
    struct pmap_registry *pmap = NULL;
    int p = 0;

    GF_ASSERT(this);

    pmap = pmap_registry_get(this);

    while (true) {
        /* coverity[DC.WEAK_CRYPTO] */
        p = (rand() % (pmap->max_port - pmap->base_port + 1)) + pmap->base_port;
        if (pmap_port_isfree(p)) {
            break;
        }
    }

    return p;
}

The issue in the above code is if the port range is exhausted it would never break/return and it will become an infinite loop.
We need to either limit the number of times the check is done or use a list/array to exit the loop if all the ports in the base_port to max_port range are checked.

The issue was noticed in the latest gluster-10 functional test on redant framework for the test case - test_reserved_port_range_for_gluster..