mgr/cephadm: check-host should not fail as hard using fqdn

Print error message instead of traceback when check-host
fails in due to host address not being found

Fixes: https://tracker.ceph.com/issues/45724
Signed-off-by: Adam King <adking@redhat.com>

diff --git a/src/pybind/mgr/cephadm/module.py b/src/pybind/mgr/cephadm/module.py
index 0f1ca5a2660608489b56eae36de8a4f8ab953327..c83d0d2e17304a8e2c9cac452c07a239e39438aa 100644 (file)
--- a/src/pybind/mgr/cephadm/module.py
+++ b/src/pybind/mgr/cephadm/module.py
@@ -844,12 +844,17 @@ class CephadmOrchestrator(orchestrator.Orchestrator, MgrModule):
         'name=addr,type=CephString,req=false',
         'Check whether we can access and manage a remote host')
     def check_host(self, host, addr=None):
-        out, err, code = self._run_cephadm(host, 'client', 'check-host',
-                                           ['--expect-hostname', host],
-                                           addr=addr,
-                                           error_ok=True, no_fsid=True)
-        if code:
-            return 1, '', ('check-host failed:\n' + '\n'.join(err))
+        try:
+            out, err, code = self._run_cephadm(host, 'client', 'check-host',
+                                               ['--expect-hostname', host],
+                                               addr=addr,
+                                               error_ok=True, no_fsid=True)
+            if code:
+                return 1, '', ('check-host failed:\n' + '\n'.join(err))
+        except OrchestratorError as e:
+            self.log.exception(f"check-host failed for '{host}'")
+            return 1, '', ('check-host failed:\n' +
+                f"Host '{host}' not found. Use 'ceph orch host ls' to see all managed hosts.")
         # if we have an outstanding health alert for this host, give the
         # serve thread a kick
         if 'CEPHADM_HOST_CHECK_FAILED' in self.health_checks:
@@ -878,7 +883,7 @@ class CephadmOrchestrator(orchestrator.Orchestrator, MgrModule):
                     self.event.set()
         return 0, '%s (%s) ok' % (host, addr), err
 
-    def _get_connection(self, host):
+    def _get_connection(self, host: str):
         """
         Setup a connection for running commands on remote host.
         """
@@ -933,6 +938,8 @@ class CephadmOrchestrator(orchestrator.Orchestrator, MgrModule):
 
         try:
             try:
+                if not addr:
+                    raise OrchestratorError("host address is empty")
                 conn, connr = self._get_connection(addr)
             except OSError as e:
                 self._reset_con(host)

diff --git a/src/pybind/mgr/cephadm/tests/test_cephadm.py b/src/pybind/mgr/cephadm/tests/test_cephadm.py
index ad7f510d990e64f00442df445d4c9610facbf886..c3938af945a19f70559d0e7d3cf4f49c8e410f58 100644 (file)
--- a/src/pybind/mgr/cephadm/tests/test_cephadm.py
+++ b/src/pybind/mgr/cephadm/tests/test_cephadm.py
@@ -579,7 +579,7 @@ class TestCephadm(object):
             _get_connection.side_effect = HostNotFound
             code, out, err = cephadm_module.check_host('test')
             assert out == ''
-            assert 'Failed to connect to test (test)' in err
+            assert "Host 'test' not found" in err
 
             out = wait(cephadm_module, cephadm_module.get_hosts())[0].to_json()
             assert out == HostSpec('test', 'test', status='Offline').to_json()