pcp
[Top] [All Lists]

Re: [pcp] per cpu utilisation from archive - pmval

To: Allan McAleavy <allan.mcaleavy@xxxxxxxxx>
Subject: Re: [pcp] per cpu utilisation from archive - pmval
From: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu, 19 May 2016 21:19:55 +1000
Cc: pcp developers <pcp@xxxxxxxxxxx>
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <CAF6XsOeRrup35qqzpbB7mKH+cOdKrPHoLJENZDobn5qPWTUzfA@xxxxxxxxxxxxxx>
References: <CAF6XsOcfikRQ-EJ9cM+iuHhpHEinvKaKDWm3oB27gMPwF2vcOw@xxxxxxxxxxxxxx> <573D0011.5090505@xxxxxxxxxxxxxxxx> <CAF6XsOeRrup35qqzpbB7mKH+cOdKrPHoLJENZDobn5qPWTUzfA@xxxxxxxxxxxxxx>
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.7.2
On 19/05/16 21:04, Allan McAleavy wrote:
Hi All

Ok so I have this now working, my pmlogger was running as follows, so
the data was in 1 min samples.

pmlogger -P -r -T24h10m -c config.default -m pmlogger_check 20160519.10.50

Changing to -t 1 gives the appropriate result. What is the best practice
for this? or should I have still been able to get similar results with
the default option? version running is pcp-3.10.9-6.el6.x86_64

Thanks for everyone's help on this.

Not sure I'm following all of this.

Adding -t 1 to the pmlogger command line would change the logging interval from 1 minute to 1 second for any metrics named in config.default with a "default" logging interval.

Depending on the contents of config.default this could mean something between no change and 60 times more data being logged.

The choice if -t for the replay tools is largely independent of the choice of the logging intervals in the archive ... most of the tools use interpolate mode where we effectively do straight line interpolation between the logged data to synthesize the values at the times the reporting tools are asking for it.

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