pcp
[Top] [All Lists]

Re: Metric for host liveness

To: Sai p Seshasayee <sseshas@xxxxxxxxxx>
Subject: Re: Metric for host liveness
From: Nathan Scott <nscott@xxxxxxxxxx>
Date: Wed, 24 Sep 2008 09:21:07 +1000
Cc: pcp@xxxxxxxxxxx
In-reply-to: <OF39E5991A.1AD4317C-ON852574CD.0046A948-852574CD.0046C58F@us.ibm.com>
References: <OF39E5991A.1AD4317C-ON852574CD.0046A948-852574CD.0046C58F@us.ibm.com>
Sender: pcp-bounce@xxxxxxxxxxx
On Tue, 2008-09-23 at 08:53 -0400, Sai p Seshasayee wrote:
> 
> Hi Nathan, 
> 
> I have installed PCP in multiple hosts. I want to know whether  a
> particular host is up or not. Is there any metric available in PCP
> which will help me 
> find whether a host is alive or not? There are over 500 metrics in PCP
> and I am finding it difficult to find correct metric. Your inputs
> would be welcome. 

If a host is not up, you will be unable to connect to pmcd on that
host - so, no metric is going to give you what you're after here I
think.

There is a pmcd_wait(1) tool which you can use to test whether pmcd
is responding on a host - its exit status will tell you whether the
connection could be established or not.

If you need this state exported as a metric for remote hosts, you
might find the pmdashping(1) agent useful - you could run pmcd_wait
for each monitored host there, and the exit status is available as
one of the shping.* metrics.

cheers.

--
Nathan


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