scylla_node: no need to _wait_no_pending_flushes

[bhalevy]

The flush api is synchronous and so it waits for
the respective flush to get done and will return
and error if flush failed for any reason and
this will propagate through nodetool flush.

_wait_no_pending_flushes may fail if there
are spontanuous flushes, and it also tends to
timeout in nodetool cfstats in debug mode,
see for example:
https://jenkins.scylladb.com/view/master/job/scylla-master/job/dtest-debug/250/testReport/alternator_ttl_tests/TestAlternatorTTL/Run_Dtest_Parallel_Cloud_Machines___FullDtest___full_split029___test_ttl_with_load_and_decommission/

../scylla/.local/lib/python3.11/site-packages/ccmlib/scylla_node.py:1320: in flush
    self._wait_no_pending_flushes(verbose=kwargs.get('verbose', True))
../scylla/.local/lib/python3.11/site-packages/ccmlib/scylla_node.py:1314: in _wait_no_pending_flushes
    if not wait_for(no_pending_flushes, timeout=wait_timeout, step=1.0):
../scylla/.local/lib/python3.11/site-packages/ccmlib/common.py:675: in wait_for
    if func():
../scylla/.local/lib/python3.11/site-packages/ccmlib/scylla_node.py:1306: in no_pending_flushes
    stdout = self.nodetool('cfstats', timeout=wait_timeout, verbose=verbose)[0]
../scylla/.local/lib/python3.11/site-packages/ccmlib/scylla_node.py:754: in nodetool
    return super().nodetool(*args, **kwargs)
../scylla/.local/lib/python3.11/site-packages/ccmlib/node.py:794: in nodetool
    stdout, stderr = p.communicate(timeout=timeout)
/usr/lib64/python3.11/subprocess.py:1209: in communicate
    stdout, stderr = self._communicate(input, endtime, timeout)
/usr/lib64/python3.11/subprocess.py:2109: in _communicate
    self._check_timeout(endtime, orig_timeout, stdout, stderr)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

self = <Popen: returncode: None args: ['/jenkins/workspace/scylla-master/dtest-debu...>
endtime = 2588.165606614, orig_timeout = 60, stdout_seq = [], stderr_seq = []
skip_check_and_raise = False

    def _check_timeout(self, endtime, orig_timeout, stdout_seq, stderr_seq,
                       skip_check_and_raise=False):
        """Convenience for checking if a timeout has expired."""
        if endtime is None:
            return
        if skip_check_and_raise or _time() > endtime:
>           raise TimeoutExpired(
                    self.args, orig_timeout,
                    output=b''.join(stdout_seq) if stdout_seq else None,
                    stderr=b''.join(stderr_seq) if stderr_seq else None)
E           subprocess.TimeoutExpired: Command '['/jenkins/workspace/scylla-master/dtest-debug/scylla/.ccm/scylla-repository/93be4c0cb0f0c53fe0eb7cd4d06ba15a7fc01d29/share/cassandra/bin/nodetool', '-h', '127.0.12.2', '-p', '7199', '-Dcom.sun.jndi.rmiURLParsing=legacy', 'cfstats']' timed out after 60 seconds

[nyh]

The commit message has a long introduction to the problem but forgets to say what you actually changed :-) But I understood it reading the trivial code.

It's worth noting that that this function wait_no_pending_flushes that you are now deleting was added seven years ago in commit b03bbb1 because, as the commit message said, "nodetool flush' is asynchronous in scylla". Supposedly this changed over the years - but I can't find the right Scylla commit where this changed. The current code indeed appears to be waiting for the flush to happen: co_await replica::database::flush_tables_on_all_shards(...).

[fruch]

LGTM