pcp
[Top] [All Lists]

Re: pmie observations

To: "Davis, Todd C" <todd.c.davis@xxxxxxxxx>
Subject: Re: pmie observations
From: Ken McDonell <kenmcd@xxxxxxxxxxxxxxxxx>
Date: Thu, 17 Oct 2002 08:25:26 +1000
Cc: pcp@xxxxxxxxxxx
In-reply-to: <29AD895CE780D511A8870002A50A666D04F9086C@hdsmsx106.hd.intel.com>
Sender: pcp-bounce@xxxxxxxxxxx
On Wed, 16 Oct 2002, Davis, Todd C wrote:

> After using pmie on Linux I have observed the following:
>
> It appears that the pmie alarm action does not work on Linux.

You're correct (this is one of those GUI areas that is hard to cover
in automated QA) ... the problem is the path to the IRIX version of
xconfirm is hard-coded.  The attached patch to src/pmie/src/act.sk
should fix it ... at least it did fix it for my simple test:

> Also the '|' action operator is ineffective since only failed forks cause an
> action to fail.

OK, I guess the issue here is that we don't capture the exit status from
the "shell" action ... all the other actions will succeed by definition.

Is this the issue?  If so, I'll have to take a look at logic here ...
seems like there is one too many level of fork() that is getting in
the way.  Do you have a simple test case?

> The UPM has the expression 'environ.temp @1 - environ.temp $0 > 2' for a
                                          I hope it says --> @0
> rising temperature test.
> This looks like a test for a falling temperature instead.

Yes the man page is incorrect (I've fixed it) ... this is indeed
falling temperature ... to get rising temperature you need

        environ.temp @0 - environ.temp @1 > 2

> Are my pmie observations correct?

Pretty much ... 8^)>

Thanks.


<Prev in Thread] Current Thread [Next in Thread>