changing the error handling when executing lsit of processes

xtreemfs_client/OSDManager.py

@@ -233,7 +233,7 @@ class OSDManager(object):
         if self.debug:
             print("starting execution of delete replica commands...")
             print(str(datetime.datetime.now()))
-        processes = div_util.run_commands(delete_replica_command_list, max_processes_delete_replica)
+        processes = div_util.run_commands(delete_replica_command_list, max_processes_delete_replica, print_errors=False)
 
         # run and repeat delete commands, until they return no error
         # (if an error is returned for another reason than that one would delete the last complete replica,
@@ -273,7 +273,7 @@ class OSDManager(object):
             if self.debug:
                 print("rerunning " + str(len(errored_deletions)) + " commands because replica could not be deleted...")
 
-            processes = div_util.run_commands(errored_deletions, max_processes_change_policy)
+            processes = div_util.run_commands(errored_deletions, max_processes_change_policy, print_errors=False)
             iterations += 1
 
         if self.debug:

xtreemfs_client/div_util.py

@@ -33,9 +33,24 @@ def command_list_to_single_string(command_list):
     return single_string
 
 
-def run_commands(commands, max_processes=200):
+def print_error(finished_process):
+    print('executing process finished with error:')
+    print('process: ')
+    print(str(finished_process[0]))
+    print('stdout:')
+    print(str(finished_process[1][0]))
+    print('stderr:')
+    print(str(finished_process[1][1]))
+
+
+'''
+execute list of commands in parallel, return list of executions returned with an error
+'''
+
+
+def run_commands(commands, max_processes=200, print_errors=True):
     running_processes = set()
-    finished_processes = []
+    errored_processes = []
     for command in commands:
         started_process = subprocess.Popen(command, stderr=subprocess.PIPE, stdout=subprocess.PIPE, shell=True)
         running_processes.add(started_process)
@@ -45,20 +60,29 @@ def run_commands(commands, max_processes=200):
             for running_process in running_processes:
                 if running_process.poll() is not None:
                     difference.add(running_process)
-                    finished_processes.append((running_process.args,
-                                               running_process.communicate(),
-                                               running_process.returncode))
+                    finished_process = (running_process.args,
+                                        running_process.communicate(),
+                                        running_process.returncode)
+                    if finished_process[2] != 0:
+                        errored_processes.append(finished_process)
+                        if print_errors:
+                            print_error(finished_process)
+
             running_processes = running_processes.difference(difference)
 
     for p in running_processes:
         p.wait()
 
     for finished_process in running_processes:
-        finished_processes.append((finished_process.args,
-                                   finished_process.communicate(),
-                                   finished_process.returncode))
-
-    return finished_processes
+        finished_process = (finished_process.args,
+                            finished_process.communicate(),
+                            finished_process.returncode)
+        if finished_process[2] != 0:
+            errored_processes.append(finished_process)
+            if print_errors:
+                print_error(finished_process)
+
+    return errored_processes
     # return list(map(lambda proc: (proc.args, proc.communicate(), proc.returncode), all_processes))