give a more useful error message if remote init fails

cylc/flow/subprocpool.py

@@ -38,6 +38,7 @@
     get_platform,
 )
 from cylc.flow.task_events_mgr import TaskJobLogsRetrieveContext
+from cylc.flow.task_proxy import TaskProxy
 from cylc.flow.wallclock import get_current_time_string
 
 _XTRIG_FUNCS: dict = {}
@@ -468,7 +469,7 @@ def _run_command_exit(
         Args:
             ctx: SubProcContext object for this task.
             callback: Function to run on command exit.
-            callback_args: Arguments to proivide to callback
+            callback_args: Arguments to provide to callback
             callback_255: Function to run if command exits with a 255
                 error - usually associated with ssh being unable to
                 contact a remote host.
@@ -486,6 +487,7 @@ def _run_callback(callback, args_=None):
 
         # If cmd is fileinstall, which uses rsync, get a platform so
         # that you can use that platform's ssh command.
+        platform_name = None
         platform = None
         if isinstance(ctx.cmd_key, TaskJobLogsRetrieveContext):
             try:
@@ -505,6 +507,15 @@ def _run_callback(callback, args_=None):
             ):
                 # the first argument is not a platform
                 platform = None
+                # Backup, get a platform name from the config:
+                for arg in callback_args:
+                    if isinstance(arg, TaskProxy):
+                        platform_name = arg.tdef.rtconfig['platform']
+                    elif (
+                        isinstance(arg, list)
+                        and isinstance(arg[0], TaskProxy)
+                    ):
+                        platform_name = arg[0].tdef.rtconfig['platform']
 
         if cls.ssh_255_fail(ctx) or cls.rsync_255_fail(ctx, platform) is True:
             # Job log retrieval passes a special object as a command key
@@ -521,7 +532,7 @@ def _run_callback(callback, args_=None):
                     ' unreachable hosts'
                     f'\n* {cmd_key} will retry if another host is available.'
                 ),
-                platform or {'name': None},
+                platform or {'name': platform_name},
                 level='warning',
             )
 

tests/unit/test_subprocpool.py

@@ -25,9 +25,12 @@
 from types import SimpleNamespace
 
 from cylc.flow import LOG
+from cylc.flow.id import Tokens
+from cylc.flow.cycling.iso8601 import ISO8601Point
 from cylc.flow.task_events_mgr import TaskJobLogsRetrieveContext
 from cylc.flow.subprocctx import SubProcContext
 from cylc.flow.subprocpool import SubProcPool, _XTRIG_FUNCS, get_func
+from cylc.flow.task_proxy import TaskProxy
 
 
 class TestSubProcPool(unittest.TestCase):
@@ -312,6 +315,32 @@ def test__run_command_exit_add_to_badhosts(mock_ctx):
     assert badhosts == {'foo', 'bar', 'mouse'}
 
 
+def test__run_command_exit_add_to_badhosts_log(caplog, mock_ctx):
+    """It gets platform name from the callback args.
+    """
+    badhosts = {'foo', 'bar'}
+    SubProcPool._run_command_exit(
+        mock_ctx(cmd=['ssh']),
+        bad_hosts=badhosts,
+        callback=lambda x, t: print(str(x)),
+        callback_args=[TaskProxy(
+            Tokens('~u/w//c/t/2'),
+            SimpleNamespace(
+                name='t', dependencies={}, sequential='',
+                external_triggers=[], xtrig_labels={},
+                outputs={
+                    'submitted': [None, None], 'submit-failed': [None, None]
+                },
+                graph_children={}, rtconfig={'platform': 'foo'}
+
+            ),
+            ISO8601Point('1990')
+        )]
+    )
+    assert 'platform: foo' in caplog.records[0].message
+    assert badhosts == {'foo', 'bar', 'mouse'}
+
+
 def test__run_command_exit_rsync_fails(mock_ctx):
     """It updates the list of badhosts
     """