FW: RRD data points - need help with graph distortion
Looks like my graphs didn't come through on the E-mail. Trying different attachment type...
From: Wood, Mike [mailto:Mike.Wood at kci1.com] Sent: Monday, November 10, 2008 9:06 AM To: hobbit at hswn.dk Subject: [hobbit] FW: RRD data points - need help with graph distortion
Hi All,
I asked this question last week, but maybe I wasn't clear enough as to what I was asking...
Management wants to avoid graph distortion on my 60 day CPU Util graph if possible. I think that more datapoints would address the problem. The first graph looks good (30 day graph). The 2nd one distorts what is really going on (60 day graph - same server). Is there any way to make the 2nd graph more granular? Thanks for looking!
Mike Wood UNIX System Administrator Kinetic Concepts Inc. 5751 NW Parkway San Antonio, TX, 78249
"CONFIDENTIALITY NOTICE: This transmission (including any accompanying attachments) is confidential, is intended only for the individual or entity named above, and is likely to contain privileged, proprietary and confidential information that is exempt from disclosure requests under applicable law. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, use of or reliance upon any of the information contained in this transmission is strictly prohibited. Any inadvertent or unauthorized disclosure shall not compromise or waive the confidentiality of this transmission or any applicable attorney-client privilege.
If you have received this transmission in error, please immediately notify us at postmaster at kci1.com."
Kinetic Concepts, Inc.
In <0D2B277DB445634E85FFD372FE7B17562DF9CE51FB at AMWPVEX01.kci.com> "Wood, Mike" <Mike.Wood at kci1.com> writes:
I asked this question last week, but maybe I wasn't clear enough as to what= I was asking...
Management wants to avoid graph distortion on my 60 day CPU Util graph if p= ossible. I think that more datapoints would address the problem. The first graph looks good (30 day graph). The 2nd one distorts what is re= ally going on (60 day graph - same server). Is there any way to make the 2= nd graph more granular? Thanks for looking!
There is, but it means changin the structure of the RRD files that Hobbit creates. And it only works for new files, not existing data.
What you need to change is the "RRA" (Round Robin Archive) definition used by Hobbit when it sets up the rrd files. Each RRA contains data that comes from the measurements, but are averaged out over progressively longer intervals. By default there are 4 sets of data:
- a 5-minute average with 576 data points. This corresponds almost directly to the individual measurements Hobbit performs, since they are also done at 5 minute intervals. It holds 576*5 minutes = 48 hours of data.
- 6 of the 5-minute averages go into the next RRA, so this is a 30-minute average. Still has 576 data points, so that holds 12 days worth of data.
- 24 of the 5-minute averages go into the third RRA, making it a 120- minute (2 hour) average. 576 data points = 48 days of data.
- 288 of the 5-minute averages go into the fourth RRA, making it a 24-hour average. With 576 data points, it goes back rougly 1.5 years.
These are defined in the Hobbit sourcecode hobbitd/do_rrd.c like this:
static char rra1[] = "RRA:AVERAGE:0.5:1:576"; static char rra2[] = "RRA:AVERAGE:0.5:6:576"; static char rra3[] = "RRA:AVERAGE:0.5:24:576"; static char rra4[] = "RRA:AVERAGE:0.5:288:576";
You can see the 1, 6, 24 and 288 number of measurements that go into one data point, and each RRA has 576 data points. Since your 30-day graph was OK - a 30-day graph uses data from the third RRA, since that is the most detailed one covering 30 days - then you could extend that one to cover the 60 days you want. 60 days with a 2-hour average requires 720 data points, so if you change the "rra3" definition to static char rra3[] = "RRA:AVERAGE:0.5:24:720"; it should be enough to store the data you need for a 60-day graph at 2-hour resolution.
You can of course do more tweaks, adding more data points just means that your rrd files will take up a bit more disk space.
After changing this, you must re-compile Hobbit (run "make clean; make; make install"), and restart Hobbit. Also, this only applies to newly created RRD files, so you will have to delete the current RRD files - so you lose your existing data.
All of this is for Hobbit 4.2.x. The coming version makes the RRD definitions configurable without having to muck about with the source code.
Regards, Henrik
On Monday 10 November 2008, Wood, Mike wrote:
Looks like my graphs didn't come through on the E-mail. Trying different attachment type...
From: Wood, Mike [mailto:Mike.Wood at kci1.com] Sent: Monday, November 10, 2008 9:06 AM To: hobbit at hswn.dk Subject: [hobbit] FW: RRD data points - need help with graph distortion
Hi All,
I asked this question last week, but maybe I wasn't clear enough as to what I was asking... - Management wants to avoid graph distortion on my 60 day CPU Util graph if possible. I think that more datapoints would address the problem. The first graph looks good (30 day graph). The 2nd one distorts what is really going on (60 day graph - same server). Is there any way to make the 2nd graph more granular? Thanks for looking! - I have a perl script to update all the rrd files:
it uses rrdtool resize <rrd file> GROW to add extra datapoints so the rrd has 5760 data points / rra (per default it has 756 data points)
it adds also MAX and MIN rra's, this is more tricky because I export the rrd to xml, change the xml info and revert back to an rrd file.
Unfortunatly, I can not share this script. But with use RRDs ; use File::Find (); you can get info about the file and find them ;)
Stef
I also remember seeing an add-on which allowed you to dump your data points into a MySQL database. This would allow you to maintain max granularity for as long as your disk space held out. If I recall, there was no modification or recompilation of the code required. (But don't quote me on that) Search the archive, and see what comes up.
Cheers Vernon
-----Original Message----- From: Stef Coene [mailto:stef.coene at docum.org] Sent: Tuesday, 11 November 2008 7:15 AM To: hobbit at hswn.dk Subject: Re: [hobbit] FW: RRD data points - need help with graph distortion
On Monday 10 November 2008, Wood, Mike wrote:
Looks like my graphs didn't come through on the E-mail. Trying different attachment type...
From: Wood, Mike [mailto:Mike.Wood at kci1.com] Sent: Monday, November 10, 2008 9:06 AM To: hobbit at hswn.dk Subject: [hobbit] FW: RRD data points - need help with graph distortion
Hi All,
I asked this question last week, but maybe I wasn't clear enough as to what I was asking... - Management wants to avoid graph distortion on my 60 day CPU Util graph if possible. I think that more datapoints would address the problem. The first graph looks good (30 day graph). The 2nd one distorts what is really going on (60 day graph - same server). Is there any way to make the 2nd graph more granular? Thanks for looking! - I have a perl script to update all the rrd files:
it uses rrdtool resize <rrd file> GROW to add extra datapoints so the rrd has 5760 data points / rra (per default it has 756 data points)
it adds also MAX and MIN rra's, this is more tricky because I export the rrd to xml, change the xml info and revert back to an rrd file.
Unfortunatly, I can not share this script. But with use RRDs ; use File::Find (); you can get info about the file and find them ;)
Stef
To unsubscribe from the hobbit list, send an e-mail to hobbit-unsubscribe at hswn.dk
NOTICE: This email and any attachments are confidential. They may contain legally privileged information or copyright material. You must not read, copy, use or disclose them without authorisation. If you are not an intended recipient, please contact us at once by return email and then delete both messages and all attachments.
participants (4)
-
henrik＠hswn.dk
-
Mike.Wood＠kci1.com
-
stef.coene＠docum.org
-
Vernon.Everett＠woodside.com.au