[PATCH] don't run test 167 if killall is not installed

[PATCH] don't run test 167 if killall is not installed

[Christoph Hellwig]

Test 167 will leave runaway fsstress processes around in case killall is
not installed and thus make all following tests fail.  This patch checks
for killall beeing installed and error out otherwise.


Signed-off-by: Christoph Hellwig <hch@lst.de>

Index: xfstests/167
===================================================================
RCS file: /cvs/xfs-cmds/xfstests/167,v
retrieving revision 1.3
diff -u -p -r1.3 167
--- xfstests/167	2 Aug 2007 16:19:56 -0000	1.3
+++ xfstests/167	15 May 2008 05:34:20 -0000
@@ -13,6 +13,7 @@ owner=dgc@sgi.com
 seq=`basename $0`
 echo "QA output created by $seq"
 
+killall="/usr/bin/killall"
 here=`pwd`
 tmp=/tmp/$$
 rm -f $seq.full
@@ -43,6 +44,8 @@ workout()
 _supported_fs xfs
 _supported_os Linux
 
+[ -x $killall ] || _notrun "$killall executable not found"
+
 _setup_testdir
 _require_scratch
 _scratch_mkfs_xfs >/dev/null 2>&1

Re: [PATCH] don't run test 167 if killall is not installed

[Tim Shimmin]

Christoph Hellwig wrote:
> Test 167 will leave runaway fsstress processes around in case killall is
> not installed and thus make all following tests fail.  This patch checks
> for killall beeing installed and error out otherwise.
> 
> 

Looks reasonable.

However:

* could use set_prog_path like we do in common.config.

* I wonder if one could use "kill -$pgid" on the process group
for fsstress instead of killall (I've never tried it :).

--Tim

> Signed-off-by: Christoph Hellwig <hch@lst.de>
> 
> Index: xfstests/167
> ===================================================================
> RCS file: /cvs/xfs-cmds/xfstests/167,v
> retrieving revision 1.3
> diff -u -p -r1.3 167
> --- xfstests/167	2 Aug 2007 16:19:56 -0000	1.3
> +++ xfstests/167	15 May 2008 05:34:20 -0000
> @@ -13,6 +13,7 @@ owner=dgc@sgi.com
>  seq=`basename $0`
>  echo "QA output created by $seq"
>  
> +killall="/usr/bin/killall"
>  here=`pwd`
>  tmp=/tmp/$$
>  rm -f $seq.full
> @@ -43,6 +44,8 @@ workout()
>  _supported_fs xfs
>  _supported_os Linux
>  
> +[ -x $killall ] || _notrun "$killall executable not found"
> +
>  _setup_testdir
>  _require_scratch
>  _scratch_mkfs_xfs >/dev/null 2>&1
> 

Re: [PATCH] don't run test 167 if killall is not installed

[David Chinner]

On Thu, May 15, 2008 at 07:39:18AM +0200, Christoph Hellwig wrote:
> Test 167 will leave runaway fsstress processes around in case killall is
> not installed and thus make all following tests fail.  This patch checks
> for killall beeing installed and error out otherwise.

Looks sane to me. I'll check it in in a minute....

Cheers,

Dave.
-- 
Dave Chinner
Principal Engineer
SGI Australian Software Group

Re: [PATCH] don't run test 167 if killall is not installed

[Christoph Hellwig]

On Thu, May 15, 2008 at 04:40:01PM +1000, Tim Shimmin wrote:
> Looks reasonable.
> 
> However:
> 
> * could use set_prog_path like we do in common.config.
> 
> * I wonder if one could use "kill -$pgid" on the process group
> for fsstress instead of killall (I've never tried it :).

Dave already commited the original version, but I'll send an update
to use set_prog_path later today.

Re: [PATCH] don't run test 167 if killall is not installed

[David Chinner]

On Thu, May 15, 2008 at 09:13:38AM +0200, Christoph Hellwig wrote:
> On Thu, May 15, 2008 at 04:40:01PM +1000, Tim Shimmin wrote:
> > Looks reasonable.
> > 
> > However:
> > 
> > * could use set_prog_path like we do in common.config.
> > 
> > * I wonder if one could use "kill -$pgid" on the process group
> > for fsstress instead of killall (I've never tried it :).
> 
> Dave already commited the original version, but I'll send an update
> to use set_prog_path later today.

Sorry, didn't see that Tim replied as well. set_prog_path is fine
by me too, but stopping every test from running because killall is
not present is a bit of overkill, isn't it?

Cheers,

Dave.
-- 
Dave Chinner
Principal Engineer
SGI Australian Software Group

Re: [PATCH] don't run test 167 if killall is not installed

[Christoph Hellwig]

On Thu, May 15, 2008 at 05:36:05PM +1000, David Chinner wrote:
> > Dave already commited the original version, but I'll send an update
> > to use set_prog_path later today.
> 
> Sorry, didn't see that Tim replied as well. set_prog_path is fine
> by me too, but stopping every test from running because killall is
> not present is a bit of overkill, isn't it?

My plan was to use it in 167 not common.config.  But I don't really care
either way.

Re: [PATCH] don't run test 167 if killall is not installed

[Timothy Shimmin]

Christoph Hellwig wrote:
> On Thu, May 15, 2008 at 05:36:05PM +1000, David Chinner wrote:
>>> Dave already commited the original version, but I'll send an update
>>> to use set_prog_path later today.
>> Sorry, didn't see that Tim replied as well. set_prog_path is fine
>> by me too, but stopping every test from running because killall is
>> not present is a bit of overkill, isn't it?
> 
> My plan was to use it in 167 not common.config.  But I don't really care
> either way.

Yeah, I was just meaning for 167 too.

--Tim