node boot: Format even with unreachable master

Use the format return code to distinguish between failure to format or
failure to report to master at the end. Then make boot run format even
if the master is unreachable, and:
* if format fails to report, wait until master is reachable, and retry
* if format fails for another reason, retry after a fixed 15s delay
* once format succeeds fully, bang

slapos/cli/boot.py

@@ -40,7 +40,7 @@ from netaddr import valid_ipv4, valid_ipv6
 from slapos.cli.command import check_root_user
 from slapos.cli.entry import SlapOSApp
 from slapos.cli.config import ConfigCommand
-from slapos.format import isGlobalScopeAddress
+from slapos.format import isGlobalScopeAddress, FormatReturn
 from slapos.util import string_to_boolean
 import argparse
 import logging
@@ -65,10 +65,7 @@ def _runBang(app):
     Launch slapos node format.
     """
     logger.info("[BOOT] Invoking slapos node bang...")
-    result = app.run(['node', 'bang', '-m', 'Reboot'])
-    if result == 1:
-      return 0
-    return 1
+    return app.run(['node', 'bang', '-m', 'Reboot'])
 
 
 def _runFormat(app):
@@ -76,10 +73,7 @@ def _runFormat(app):
     Launch slapos node format.
     """
     logger.info("[BOOT] Invoking slapos node format...")
-    result = app.run(['node', 'format', '--now', '--verbose'])
-    if result == 1:
-      return 0
-    return 1
+    return app.run(['node', 'format', '--now', '--verbose'])
 
 
 def _ping(hostname):
@@ -139,6 +133,16 @@ def _ping_hostname(hostname):
       is_ready = _ping6(hostname)
 
 
+def _ping_master(master_hostname):
+  if valid_ipv4(master_hostname):
+    _test_ping(master_hostname)
+  elif valid_ipv6(master_hostname):
+    _test_ping6(master_hostname)
+  else:
+    # hostname
+    _ping_hostname(master_hostname)
+
+
 def _waitIpv6Ready(ipv6_interface):
   """
     test if ipv6 is ready on ipv6_interface
@@ -154,6 +158,7 @@ def _waitIpv6Ready(ipv6_interface):
           "try again in 5 seconds...", ipv6_interface)
     sleep(5)
 
+
 class BootCommand(ConfigCommand):
     """
     Test network and invoke simple format and bang (Use on Linux startup)
@@ -196,23 +201,29 @@ class BootCommand(ConfigCommand):
         if ipv6_interface is not None:
             _waitIpv6Ready(ipv6_interface)
 
-        # Check that node can ping master
-        if valid_ipv4(master_hostname):
-            _test_ping(master_hostname)
-        elif valid_ipv6(master_hostname):
-            _test_ping6(master_hostname)
-        else:
-            # hostname
-            _ping_hostname(master_hostname)
-
         app = SlapOSApp()
+        while True:
             # Make sure slapos node format returns ok
-        while not _runFormat(app):
+            result = _runFormat(app)
+
+            if result == FormatReturn.FAILURE:
                 logger.error("[BOOT] Fail to format, try again in 15 seconds...")
                 sleep(15)
+                continue
+
+            if result == FormatReturn.OFFLINE_SUCCESS:
+                logger.error(
+                    "[BOOT] Fail to post format information"
+                    ", try again when connection to master is up..."
+                )
+                sleep(15)
+                _ping_master(master_hostname)
+                continue
+
+            break
 
         # Make sure slapos node bang returns ok
-        while not _runBang(app):
+        while _runBang(app):
             logger.error("[BOOT] Fail to bang, try again in 15 seconds...")
             sleep(15)
 

slapos/cli/format.py

@@ -125,4 +125,4 @@ class FormatCommand(ConfigCommand):
 
         tracing_monkeypatch(conf)
 
-        do_format(conf=conf)
+        return do_format(conf=conf)

slapos/format.py

@@ -30,6 +30,7 @@
 
 from six.moves import configparser
 import distro
+import enum
 import errno
 import fcntl
 import grp
@@ -68,6 +69,12 @@ from slapos import version
 from slapos import manager as slapmanager
 
 
+class FormatReturn(enum.IntEnum):
+  SUCCESS = 0
+  FAILURE = 1
+  OFFLINE_SUCCESS = 2
+
+
 logger = logging.getLogger("slapos.format")
 
 
@@ -1578,6 +1585,7 @@ def random_delay(conf):
 
 
 def do_format(conf):
+  try:
     random_delay(conf)
 
     if conf.input_definition_file:
@@ -1609,8 +1617,18 @@ def do_format(conf):
                     path_to_json=conf.computer_json,
                     logger=conf.logger)
     conf.logger.info('Posting information to %r' % conf.master_url)
+    try:
       computer.send(conf)
+      return FormatReturn.SUCCESS
+    except Exception:
+      conf.logger.exception('failed to transfer information to %r' % conf.master_url)
+      return FormatReturn.OFFLINE_SUCCESS
+    finally:
       conf.logger.info('slapos successfully prepared the computer.')
+  except Exception:
+    conf.logger.exception('slapos failed to prepare the computer.')
+    return FormatReturn.FAILURE
+
 
 
 class FormatConfig(object):

slapos/tests/test_cli.py

@@ -436,8 +436,9 @@ class TestCliBoot(CliMixin):
 
       # run slapos node boot
       app = slapos.cli.entry.SlapOSApp()
+      fake = mock.Mock(return_value=mock.Mock(**{'run.return_value': 0}))
       with patch('slapos.cli.boot.check_root_user', return_value=True) as check_root_user,\
-          patch('slapos.cli.boot.SlapOSApp') as SlapOSApp,\
+          patch('slapos.cli.boot.SlapOSApp', new=fake) as SlapOSApp,\
           patch('slapos.cli.boot.ConfigCommand.config_path', return_value=slapos_conf.name), \
           patch(
               'slapos.cli.boot.netifaces.ifaddresses',
@@ -477,8 +478,8 @@ class TestCliBoot(CliMixin):
         patch('slapos.cli.format.check_root_user', return_value=True),\
         patch('slapos.cli.format.logging.FileHandler', return_value=logging.NullHandler()),\
         patch('slapos.cli.bang.check_root_user', return_value=True),\
-        patch('slapos.cli.format.do_format', side_effect=[Exception, Exception, None]) as do_format,\
-        patch('slapos.cli.bang.do_bang', side_effect=[Exception, Exception, None]) as do_bang:
+        patch('slapos.cli.format.do_format', side_effect=[Exception, Exception, 0]) as do_format,\
+        patch('slapos.cli.bang.do_bang', side_effect=[Exception, Exception, 0]) as do_bang:
 
       app.run(('node', 'boot'))