pcp
[Top] [All Lists]

Re: [pcp] pmlogger_merge failing on Centos 6.4

To: chandana@xxxxxxxxxxxxx
Subject: Re: [pcp] pmlogger_merge failing on Centos 6.4
From: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon, 24 Jun 2013 18:53:27 -0400 (EDT)
Cc: pcp@xxxxxxxxxxx
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <51C8A3F6.2090103@xxxxxxxxxxxxx>
References: <mailman.1.1372093201.2271.pcp@xxxxxxxxxxx> <51C8A3F6.2090103@xxxxxxxxxxxxx>
Reply-to: Nathan Scott <nathans@xxxxxxxxxx>
Thread-index: YX8bpyfXaMmXbcpv3SiaOb4gBuMZHQ==
Thread-topic: pmlogger_merge failing on Centos 6.4
Hi,

----- Original Message -----
> proc metrics are being logged.
> 
> This is short term, while we are watching a situation
> 
> >   ... this seems excessive, _unless_
> > the proc metrics are being logged also (big indom,
> > changes every fetch, huge meta data file and we load
> > that all into memory at the archive open time).
> 

BTW, a longer-term solution here would be to use a cgroup
for reducing the process indom and reigning in its huge
variation.  Requires some coding in cgroup_procs_fetch()
in src/pmdas/linux_proc/cgroups.c though (commented block
marked with TODO in there).

cheers.

--
Nathan

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