Fix async to use the new argfiles method (wrapping brain around rock, really…

Fix async to use the new argfiles method (wrapping brain around rock, really must write module development guide)

lib/ansible/runner.py

@@ -40,6 +40,9 @@ import StringIO
 # FIXME: stop importing *, use as utils/errors
 from ansible.utils import *
 from ansible.errors import *
+
+# should be True except in debug
+CLEANUP_FILES = True
     
 ################################################
 
@@ -221,7 +224,8 @@ class Runner(object):
         for filename in files:
             if not filename.startswith('/tmp/'):
                 raise Exception("not going to happen")
-            self._exec_command(conn, "rm -rf %s" % filename)
+            if CLEANUP_FILES:
+                self._exec_command(conn, "rm -rf %s" % filename)
 
     # *****************************************************
 
@@ -255,13 +259,15 @@ class Runner(object):
         args_fo.flush()
         args_fo.close()
         args_remote = os.path.join(tmp, 'arguments')
-        self._transfer_file(conn, args_file, 'arguments')
-        os.unlink(args_file)
+        self._transfer_file(conn, args_file, args_remote)
+        if CLEANUP_FILES:
+            os.unlink(args_file)
         return args_remote
 
     # *****************************************************
 
-    def _execute_module(self, conn, tmp, remote_module_path, module_args):
+    def _execute_module(self, conn, tmp, remote_module_path, module_args, 
+        async_jid=None, async_module=None, async_limit=None):
         ''' 
         runs a module that has already been transferred, but first
         modifies the command using setup_cache variables (see playbook)
@@ -286,7 +292,11 @@ class Runner(object):
         args = template.render(inject_vars)
 
         argsfile = self._transfer_argsfile(conn, tmp, args)
-        cmd = "%s %s" % (remote_module_path, 'arguments')
+        if async_jid is None:
+            cmd = "%s %s" % (remote_module_path, argsfile)
+        else:
+            args = [str(x) for x in [remote_module_path, async_jid, async_limit, async_module, argsfile]]
+            cmd = " ".join(args)
         result = self._exec_command(conn, cmd)
         return result
 
@@ -324,10 +334,11 @@ class Runner(object):
 
         async  = self._transfer_module(conn, tmp, 'async_wrapper')
         module = self._transfer_module(conn, tmp, self.module_name)
-        new_args = [] 
-        new_args = [ self.generated_jid, self.background, module ]
-        new_args.extend(self.module_args)
-        result = self._execute_module(conn, tmp, async, new_args)
+        result = self._execute_module(conn, tmp, async, self.module_args,
+           async_module=module, 
+           async_jid=self.generated_jid, 
+           async_limit=self.background
+        )
         return self._return_from_module(conn, host, result)
 
     # *****************************************************
@@ -454,8 +465,7 @@ class Runner(object):
         msg = '%s: %s' % (self.module_name, cmd)
         self.remote_log(conn, msg)
         stdin, stdout, stderr = conn.exec_command(cmd)
-        results = "\n".join(stdout.readlines())
-        return results
+        return "\n".join(stdout.readlines())
 
     # *****************************************************
 

library/async_status

@@ -30,12 +30,12 @@ import datetime
 import traceback
 
 # ===========================================
-# convert arguments of form a=b c=d
-# to a dictionary
-# FIXME: make more idiomatic
 
-args = " ".join(sys.argv[1:])
-items = shlex.split(args)
+# FIXME: better error handling
+
+argsfile = sys.argv[1]
+items = shlex.split(file(argsfile).read())
+
 params = {}
 for x in items:
     (k, v) = x.split("=")

library/async_wrapper

@@ -66,16 +66,15 @@ def daemonize_self():
 if len(sys.argv) < 3:
     print json.dumps({
         "failed" : True,
-        "msg"    : "usage: async_wrapper <jid> <module_script> <time_limit> <args>.  Humans, do not call directly!"
+        "msg"    : "usage: async_wrapper <jid> <time_limit> <modulescript> <argsfile>.  Humans, do not call directly!"
     })
     sys.exit(1)
 
 jid = sys.argv[1]
 time_limit = sys.argv[2]
 wrapped_module = sys.argv[3]
-args = sys.argv[4:]
-
-cmd = "%s %s" % (wrapped_module, " ".join(args))
+argsfile = sys.argv[4]
+cmd = "%s %s" % (wrapped_module, argsfile)
 
 # setup logging directory
 logdir = os.path.expanduser("~/.ansible_async")
@@ -92,20 +91,20 @@ if not os.path.exists(logdir):
 
 def _run_command(wrapped_cmd, jid, log_path):
 
-    print "RUNNING: %s" % wrapped_cmd
     logfile = open(log_path, "w")
     logfile.write(json.dumps({ "started" : 1, "ansible_job_id" : jid }))
     logfile.close()
     logfile = open(log_path, "w")
     result = {}
-
+  
+    outdata = ''
     try:
         cmd = shlex.split(wrapped_cmd)
         script = subprocess.Popen(cmd, shell=False, 
             stdin=None, stdout=logfile, stderr=logfile)
         script.communicate()
-        #result = json.loads(out)
-        result = json.loads(file(log_path).read())    
+        outdata = file(log_path).read()
+        result = json.loads(outdata)
 
     except (OSError, IOError), e:
         result = {
@@ -119,6 +118,7 @@ def _run_command(wrapped_cmd, jid, log_path):
         result = {
             "failed" : 1,
             "cmd" : wrapped_cmd,
+            "data" : outdata, # temporary debug only
             "msg" : traceback.format_exc()
         }   
         result['ansible_job_id'] = jid

library/command

@@ -30,32 +30,10 @@ import traceback
 import shlex
 import os
 
-if len(sys.argv) == 1:
-    print json.dumps({
-        "failed" : True,
-        "msg"    : "the command module requires arguments (-a)"
-    })
-    sys.exit(1)
-
 argfile = sys.argv[1]
-if not os.path.exists(argfile):
-    print json.dumps({
-        "failed" : True,
-        "msg"    : "Argument file not found"
-    })
-    sys.exit(1)
-
 args = open(argfile, 'r').read()
 args = shlex.split(args)
 
-if not len(args):
-    print json.dumps({
-        "failed" : True,
-        "msg"    : "the command module requires arguments (-a)"
-    })
-    sys.exit(1)
-
-
 startd = datetime.datetime.now()
 
 try:

test/TestRunner.py

@@ -159,14 +159,18 @@ class TestRunner(unittest.TestCase):
    def test_async(self):
        # test async launch and job status
        # of any particular module
+       print "firing command..."
        result = self._run('command', [ "/bin/sleep", "3" ], background=20)
+       print "back..."
        assert 'ansible_job_id' in result
        assert 'started' in result
        jid = result['ansible_job_id']
        # no real chance of this op taking a while, but whatever
        time.sleep(5)
        # CLI will abstract this, but this is how it works internally
+       print "checking status..."
        result = self._run('async_status', [ "jid=%s" % jid ])
+       print "back..."
        # TODO: would be nice to have tests for supervisory process
        # killing job after X seconds
        assert 'finished' in result