pcp
[Top] [All Lists]

Re: [pcp] PCP libvirt PMDA

To: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Subject: Re: [pcp] PCP libvirt PMDA
From: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue, 26 Jul 2016 23:53:00 -0400 (EDT)
Cc: pcp@xxxxxxxxxxx
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <57967863.8090605@xxxxxxxxxxxxxxxx>
References: <1fa58d82-ac73-7747-c58d-acf880bc2155@xxxxxxxxxx> <fc9e92aa-355e-6cc1-7586-599847cf6cbe@xxxxxxxxxx> <1473242169.7374530.1468971331935.JavaMail.zimbra@xxxxxxxxxx> <810341356.7539206.1469047436415.JavaMail.zimbra@xxxxxxxxxx> <6030869a-395c-6d2f-9277-9d3d307c6bd1@xxxxxxxxxx> <1905833000.8482698.1469408550782.JavaMail.zimbra@xxxxxxxxxx> <etPan.57961a12.4a39409f.2ce@xxxxxxxxxxx> <57967863.8090605@xxxxxxxxxxxxxxxx>
Reply-to: Nathan Scott <nathans@xxxxxxxxxx>
Thread-index: aQ9DtzGMNu+3EJUZZbxgnBOrrz+sEg==
Thread-topic: PCP libvirt PMDA

----- Original Message -----
> On 25/07/16 23:54, Martins Innus wrote:
> > ... We havenât started logging any nodes with this pmda, just
> > starting to figure out what is useful, but whatever the final choice
> > is, it would be ideal if pmlogger would automatically pickup any new
> > devices.  So if the pmlogger config is libvirt.* , and the VMs
> > running at pmlogger start time each have only one network device and
> > one blockdevice, if a VM is then started with multiple of each, those
> > start getting logged as well.
> 
> I haven't been following this thread in detail, so this comment may not
> be relevant, but ...
> 

(it's very relevant)

> If the PMDA is using dynamic metric names (not dynamic instances) for
> this data, then pmlogger will only log the metric _names_ that exist at
> the time pmlogger starts ... instances can come and go, but the PMNS
> will not be re-explored by pmlogger once it has parsed the configuration
> file.

Right.  Similar to the pmie problem I mentioned originally, where client
handling of instances is inherently far more dynamic than metrics.

cheers.

--
Nathan

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