Merge pull request #57 from ClogenyTechnologies/kd/processgrp-fix

aix getpgid fails when called from parent and parent and child have different session

2 files changed, 18 insertions, 12 deletions

diff --git a/lib/mixlib/shellout/unix.rb b/lib/mixlib/shellout/unix.rb
index cd9790a..2bde64a 100644
--- a/lib/mixlib/shellout/unix.rb
+++ b/lib/mixlib/shellout/unix.rb
@@ -280,7 +280,10 @@ module Mixlib
           # support the "ONESHOT" optimization (where sh -c does exec without
           # forking). To support cleaning up all the children, we need to
           # ensure they're in a unique process group.
-          Process.setsid
+          # We cannot use setsid here since getpgid fails on AIX with EPERM 
+          # when parent and child have different sessions and the parent tries to get the process group,
+          # hence we just create a new process group, and have the same session.
+          Process.setpgrp
 
           configure_subprocess_file_descriptors
 
diff --git a/spec/mixlib/shellout_spec.rb b/spec/mixlib/shellout_spec.rb
index f7625e1..cb0a33d 100644
--- a/spec/mixlib/shellout_spec.rb
+++ b/spec/mixlib/shellout_spec.rb
@@ -935,7 +935,7 @@ describe Mixlib::ShellOut do
         let(:cmd) { [ 'exit' ] }
 
         it "handles ESRCH from getpgid of a zombie", :unix_only do
-          Process.stub(:setsid) { exit!(4) }
+          Process.stub(:setpgrp) { exit!(4) }
 
           # there is a small race condition here if the child doesn't get
           # scheduled and call exit! before the parent can call getpgid, so run
@@ -1068,16 +1068,19 @@ describe Mixlib::ShellOut do
               shell_cmd.stdout.should include("got term in child")
               shell_cmd.stdout.should include("got term in grandchild")
 
-              Process.kill(:INT, child_pgid) # should raise ESRCH
-
-              # Debug the failure:
-              puts "child pgid=#{child_pgid.inspect}"
-              Process.wait
-              puts "collected process: #{$?.inspect}"
-              puts "initial process listing:\n#{initial_process_listing}"
-              puts "current process listing:"
-              puts `ps -j`
-              raise "Failed to kill all expected processes"
+              kill_return_val = Process.kill(:INT, child_pgid) # should raise ESRCH
+              # AIX - kill returns code > 0 for error, where as other platforms return -1. Ruby code signal.c treats < 0 as error and raises exception and hence fails on AIX. So we check the return code for assertions since ruby wont raise an error here.
+
+              if(kill_return_val == 0)
+                # Debug the failure:
+                puts "child pgid=#{child_pgid.inspect}"
+                Process.wait
+                puts "collected process: #{$?.inspect}"
+                puts "initial process listing:\n#{initial_process_listing}"
+                puts "current process listing:"
+                puts `ps -j`
+                raise "Failed to kill all expected processes"
+              end
             rescue Errno::ESRCH
               # this is what we want
             end