[Gluster-devel] Jenkins regression for release-3.7 messed up?

[Ravishankar N]

'brick_up_status' is used by the following .ts in release-3.7

[root at ravi1 glusterfs]# git grep -w brick_up_status

tests/bugs/bitrot/bug-1288490.t:EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" 
brick_up_status $V0 $H0 $B0/brick0
tests/bugs/bitrot/bug-1288490.t:EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" 
brick_up_status $V0 $H0 $B0/brick1
tests/bugs/glusterd/bug-1225716-brick-online-validation-remove-brick.t:EXPECT_WITHIN 
$PROCESS_UP_TIMEOUT "Y" brick_up_status $V0 $H0 $B0/${V0}1
tests/bugs/glusterd/bug-857330/normal.t:EXPECT_WITHIN 
$PROCESS_UP_TIMEOUT "Y" brick_up_status $V0 $H0 $B0/${V0}3
tests/bugs/glusterd/bug-857330/xml.t:EXPECT_WITHIN $PROCESS_UP_TIMEOUT 
"Y" brick_up_status $V0 $H0 $B0/${V0}3

There seems to be a bug in this function. (It is another matter that the 
function is different in master but let us ignore that for now).  So all 
these tests should fail on release-3.7 and they do fail on my 
machine.*But for some reason, they succeed on jenkins. Why is that? They 
are not in bad_tests on 3.7 either.**
*
This fixes the function:
diff --git a/tests/volume.rc b/tests/volume.rc
index 9bd9eca..6040c5f 100644
--- a/tests/volume.rc
+++ b/tests/volume.rc
@@ -24,7 +24,7 @@ function brick_up_status {
          local host=$2
          local brick=$3
          brick_pid=$(get_brick_pid $vol $host $brick)
-        gluster volume status | grep $brick_pid | awk '{print $4}'
+        gluster volume status | grep $brick_pid | awk '{print $5}'
  }

and all the tests pass with the fix on my machine. I had send the fix as 
a part of a patch[1] and it *fails* on jenkins.Why?

Thanks,
Ravi


[1] http://review.gluster.org/#/c/13609/2



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20160305/2100331c/attachment.html>