| Author |
Message |
|
graysky
|
 Posted: Sat Aug 23, 2008 10:25 am |
|
|
Joined: Wed Dec 10, 2003 8:31 pm
Posts: 1996
Location:
/dev/null
|
Can someone help me create an rrd monitor for my GPU temp? The key data gathering script is simply:
Code: /usr/bin/nvclock -T | grep GPU | cut -c 21-22
Which on my machine will return the GPU temp as a two digit number.
I can't for the life of me adapt one of my rrd_x.pl scripts to work with this. I tried using my rrd_HDDTemp.pl and making what I thought were the correct changes. Anyway, I don't know perl 
Can anyone with some perl skills have a crack at it for me?
_________________
Retired KM user (R4 - R6.04); friend to LH users.
Last edited by graysky on Sun Aug 24, 2008 11:59 pm, edited 2 times in total.
|
|
| Top |
|
 |
|
graysky
|
Posted: Sun Aug 24, 2008 2:33 pm |
|
|
Joined: Wed Dec 10, 2003 8:31 pm
Posts: 1996
Location:
/dev/null
|
I think I got it!
I adapted the rrd_MBtemp.pl to work for GPU core temp. My only problem with it now is that I can't remove the green CPU Temp portion from the graph without causing errors when the script runs.
If I comment out any of the mbt2 lines, I get errors. Here is the modified script: Code: #!/usr/bin/perl
#
# rrd_GPUtemp.pl
# GPU Core temp for KnoppMyth
#
########################################################################
# Configuration:
my $dbf = 'GPUtemp';
my $configfile = '/etc/rrd.config';
########################################################################
use RRDs;
if (! -d "/myth") { $configfile = "./D_rrd.config"; } # DEBUG
do $configfile;
sub create {
# $_[0] = filename
if (! -e "$log/$_[0].rrd") {
print "Create db for $_[0] => $log/$_[0].rrd\n";
RRDs::create( "$log/$_[0].rrd", "-s 300",
"DS:mbt1:GAUGE:600:0:U",
"DS:mbt2:GAUGE:600:0:U",
"DS:mbt3:GAUGE:600:0:U",
"RRA:AVERAGE:0.5:1:576",
"RRA:AVERAGE:0.5:6:672",
"RRA:AVERAGE:0.5:24:732",
"RRA:AVERAGE:0.5:144:1460");
$ERROR = RRDs::error;
print "Error: RRDs::create failed for '$_[0]' : $ERROR\n" if $ERROR;
}
}
#my ($mbt1, $mbt2);
my $mbt1;
sub gather {
$mbt1 = `/usr/bin/nvclock -T | grep GPU | cut -c 21-22`;
# $mbt2 = `$MBT_prog 2`;
# $mbt3 = `$MBT_prog 3`;
$mbt1 =~ s/[\n ]//g;
# $mbt2 =~ s/[\n ]//g;
# $mbt3 =~ s/[\n ]//g;
print "$dbf: gpu $mbt1, °C\n";
# The motherboard sensor occasionally returns nonsense values.
# This should keep the spurious peaks from roaching the graph...
# if ($mbt1 > 100.0) { $mbt1 = $mbt2 }
# print "$dbf: motherboard $mbt1, CPU $mbt2, case $mbt3, °C\n";
}
sub update {
# $_[0] = filename
RRDs::update( "$log/$_[0].rrd", "-t",
"mbt1",
"N:$mbt1");
$ERROR = RRDs::error;
print "Error: RRDs::update for '$_[0]' : $ERROR\n" if $ERROR;
}
sub graph {
# $_[0] = time interval (ie: day...)
# $_[1] = filename suffix.
RRDs::graph( "$png/$dbf-$_[1].png", "-s -1$_[0]", "-aPNG",
"-w $Gwd", "-h $Ght", "-E", "-l 20", "-M",
"--color", "SHADEB$color",
"--color", "SHADEA$color",
"--color", "BACK$color",
"-t nVidia GeForce 6200 Core temperature degrees C :: $_[0]",
"DEF:mbt1=$log/$dbf.rrd:mbt1:AVERAGE",
"DEF:mbt2=$log/$dbf.rrd:mbt2:AVERAGE",
"LINE1:mbt1$color_mbt1:GPU Core temp\\:",
"GPRINT:mbt1:MIN:Minimum\\: % 5.1lf",
"GPRINT:mbt1:MAX:Maximum\\: % 5.1lf",
"GPRINT:mbt1:AVERAGE:Average\\: % 5.1lf",
"GPRINT:mbt1:LAST:Current\\: % 5.1lf °C\\j",
"LINE1:mbt2$color_mbt2:CPU temp\\:",
"GPRINT:mbt2:MIN:Minimum\\: % 5.1lf",
"GPRINT:mbt2:MAX:Maximum\\: % 5.1lf",
"GPRINT:mbt2:AVERAGE:Average\\: % 5.1lf",
"GPRINT:mbt2:LAST:Current\\: % 5.1lf °C\\j");
$ERROR = RRDs::error;
print "Error: RRDs::graph failed for '$_[0]' : $ERROR\n" if $ERROR;
}
########################################################################
create "$dbf";
gather;
update "$dbf";
graph '2hour', '12hourly';
graph 'day', 'Daily';
graph 'week', 'Weekly';
graph 'month', 'Monthly';
graph 'year', 'Yearly';
########################################################################
# vim: sw=4 ts=8
# End
If anyone can tell me what I need to do to it to remove the green CPU Temp, I'd really appreciate it
_________________
Retired KM user (R4 - R6.04); friend to LH users.
Last edited by graysky on Sun Aug 24, 2008 4:22 pm, edited 1 time in total.
|
|
| Top |
|
|
|
md10md
|
Posted: Sun Aug 24, 2008 3:16 pm |
|
|
Joined: Sat Jun 25, 2005 7:39 pm
Posts: 162
|
|
Great work! I've added it to my config. Thanks again.
|
|
| Top |
|
|
|
graysky
|
Posted: Sun Aug 24, 2008 4:23 pm |
|
|
Joined: Wed Dec 10, 2003 8:31 pm
Posts: 1996
Location:
/dev/null
|
Hehe, is it working? You might have rm -rf .nvclock and then touch .nvclock in your /home/mythtv for it to work. It worked out-of-the-box on one machine, but I had to do as I described above on two other machines.
It's also pretty crude and not very elegant at the moment... I'm still hoping someone out there will help out by cleaning it up
_________________
Retired KM user (R4 - R6.04); friend to LH users.
Last edited by graysky on Sun Aug 24, 2008 4:27 pm, edited 3 times in total.
|
|
| Top |
|
|
|
md10md
|
Posted: Sun Aug 24, 2008 4:25 pm |
|
|
Joined: Sat Jun 25, 2005 7:39 pm
Posts: 162
|
|
Yeah, worked OOTB for me. I've used nvclock in the past though and I think I remember having to mess around with that file. When (If) I get around to it I'll try changing the file to see if I can get that defunct CPU temp out of their.
|
|
| Top |
|
|
|
graysky
|
Posted: Sun Aug 24, 2008 4:28 pm |
|
|
Joined: Wed Dec 10, 2003 8:31 pm
Posts: 1996
Location:
/dev/null
|
Cool man, I'm always glad to share with the community.
You do know to edit your /usr/local/bin/rrd_Config.sh and add a line for the GPUTemp right? That way, the cgi script will make your GPUTemp.html automatically.
Here is the relevant section of mine for your reference:
Code: # Establish the checklist:
TYPARY=( \
Net "Network In/Out Traffic" on \
Disk "Disk reads/writes" on \
Count "Count of files in /myth/tv" on \
MythFS "/myth Space & Inode usage" on \
HDtemp "Disk temperature" on \
GPUtemp "GPU temperature" off \
MBtemp "MB and CPU temperature" off \
MBfan "MB fan speeds" off \
UPS "UPS information" off \
_________________
Retired KM user (R4 - R6.04); friend to LH users.
|
|
| Top |
|
|
|
graysky
|
Posted: Sun Aug 24, 2008 11:05 pm |
|
|
Joined: Wed Dec 10, 2003 8:31 pm
Posts: 1996
Location:
/dev/null
|
Figured it out! It's all about the grammar of the perl script.
Original code (relevant section): Code: sub graph {
# $_[0] = time interval (ie: day...)
# $_[1] = filename suffix.
RRDs::graph( "$png/$dbf-$_[1].png", "-s -1$_[0]", "-aPNG",
"-w $Gwd", "-h $Ght", "-E", "-l 20", "-M",
"--color", "SHADEB$color",
"--color", "SHADEA$color",
"--color", "BACK$color",
"-t nVidia GeForce 6200 Core temperature degrees C :: $_[0]",
"DEF:mbt1=$log/$dbf.rrd:mbt1:AVERAGE",
"DEF:mbt2=$log/$dbf.rrd:mbt2:AVERAGE",
"LINE1:mbt1$color_mbt1:GPU Core temp\\:",
"GPRINT:mbt1:MIN:Minimum\\: % 5.1lf",
"GPRINT:mbt1:MAX:Maximum\\: % 5.1lf",
"GPRINT:mbt1:AVERAGE:Average\\: % 5.1lf",
"GPRINT:mbt1:LAST:Current\\: % 5.1lf °C\\j",
"LINE1:mbt2$color_mbt2:CPU temp\\:",
"GPRINT:mbt2:MIN:Minimum\\: % 5.1lf",
"GPRINT:mbt2:MAX:Maximum\\: % 5.1lf",
"GPRINT:mbt2:AVERAGE:Average\\: % 5.1lf",
"GPRINT:mbt2:LAST:Current\\: % 5.1lf °C\\j"); Functional code (relevant section): Code: sub graph {
# $_[0] = time interval (ie: day...)
# $_[1] = filename suffix.
RRDs::graph( "$png/$dbf-$_[1].png", "-s -1$_[0]", "-aPNG",
"-w $Gwd", "-h $Ght", "-E", "-l 40", "-r", "-v temp. (°C)", "-M",
"--color", "SHADEB$color",
"--color", "SHADEA$color",
"--color", "BACK$color",
"-t nVidia GeForce 6200 Core temperature :: $_[0]",
"DEF:mbt1=$log/$dbf.rrd:mbt1:AVERAGE",
# "DEF:mbt2=$log/$dbf.rrd:mbt2:AVERAGE",
"LINE1:mbt1$color_mbt1:GPU Core temp\\:",
"GPRINT:mbt1:MIN:Minimum\\: % 5.1lf",
"GPRINT:mbt1:MAX:Maximum\\: % 5.1lf",
"GPRINT:mbt1:AVERAGE:Average\\: % 5.1lf",
"GPRINT:mbt1:LAST:Current\\: % 5.1lf °C\\j");
# "LINE1:mbt2$color_mbt2:CPU temp\\:",
# "GPRINT:mbt2:MIN:Minimum\\: % 5.1lf",
# "GPRINT:mbt2:MAX:Maximum\\: % 5.1lf",
# "GPRINT:mbt2:AVERAGE:Average\\: % 5.1lf",
# "GPRINT:mbt2:LAST:Current\\: % 5.1lf °C\\j");
See the difference? No? It was all about the position of the ending ) mark. You have to move it up to the the end of the mbt1 lines and end with a ; (see the "GPRINT:mbt1:LAST:Current\\: % 5.1lf °C\\j"); line).
Anyway, now it works just fine:
Before
After
Oh, the other thing I did was to add the following to the line that defines the graph:
"-l 40",
"-r",
"-v temp. (°C)",
The "-l 40" limits to lower axis to 40 when you invoke the "-r" switch that rigidly enforces the limit. The "-v temp. (°C)" switch adds the text to the vertical axis. BTW, you get the ° sign by typing alt+0176. For more, see the rrdgraph docs.
_________________
Retired KM user (R4 - R6.04); friend to LH users.
|
|
| Top |
|
|
|
graysky
|
Posted: Sun Aug 24, 2008 11:58 pm |
|
|
Joined: Wed Dec 10, 2003 8:31 pm
Posts: 1996
Location:
/dev/null
|
I also combined the video card and hdd temps into one graph. Why not add the mb and cpu temp? I wrote this for a machine that's from 1997 and doesn't have mb and cpu logging!
If you're interested, here is the modified /usr/local/bin/rrd_GPUtemp.pl:
Code: #!/usr/bin/perl
#
# rrd_MBtemp.pl
# Motherboard Temperature data collection routine for KnoppMyth
#
########################################################################
# Configuration:
my $dbf = 'GPUtemp';
my $configfile = '/etc/rrd.config';
########################################################################
use RRDs;
if (! -d "/myth") { $configfile = "./D_rrd.config"; } # DEBUG
do $configfile;
sub create {
# $_[0] = filename
if (! -e "$log/$_[0].rrd") {
print "Create db for $_[0] => $log/$_[0].rrd\n";
RRDs::create( "$log/$_[0].rrd", "-s 300",
"DS:mbt1:GAUGE:600:0:U",
"DS:mbt2:GAUGE:600:0:U",
"DS:mbt3:GAUGE:600:0:U",
"RRA:AVERAGE:0.5:1:576",
"RRA:AVERAGE:0.5:6:672",
"RRA:AVERAGE:0.5:24:732",
"RRA:AVERAGE:0.5:144:1460");
$ERROR = RRDs::error;
print "Error: RRDs::create failed for '$_[0]' : $ERROR\n" if $ERROR;
}
}
my ($mbt1, $mbt2);
sub gather {
$mbt1 = `/usr/bin/nvclock -T | grep GPU | cut -c 21-22`;
$mbt2 = `/usr/sbin/hddtemp -n /dev/hda`;
# $mbt3 = `$MBT_prog 3`;
$mbt1 =~ s/[\n ]//g;
$mbt2 =~ s/[\n ]//g;
# $mbt3 =~ s/[\n ]//g;
print "$dbf: gpu temp $mbt1, hdd temp $mbt2, °C\n";
# The motherboard sensor occasionally returns nonsense values.
# This should keep the spurious peaks from roaching the graph...
if ($mbt1 > 100.0) { $mbt1 = $mbt2 }
# print "$dbf: GPU Core temp $mbt1, $mbt2, HDD temp $mbt3, °C\n";
print "$dbf: gpu temp $mbt1, hdd temp $mbt2, °C\n";
}
sub update {
# $_[0] = filename
RRDs::update( "$log/$_[0].rrd", "-t",
"mbt1:mbt2",
"N:$mbt1:$mbt2");
$ERROR = RRDs::error;
print "Error: RRDs::update for '$_[0]' : $ERROR\n" if $ERROR;
}
sub graph {
# $_[0] = time interval (ie: day...)
# $_[1] = filename suffix.
RRDs::graph( "$png/$dbf-$_[1].png", "-s -1$_[0]", "-aPNG",
"-w $Gwd", "-h $Ght", "-E", "-A",
"--color", "SHADEB$color",
"--color", "SHADEA$color",
"--color", "BACK$color",
"-t Video Card and HDD Core Temperature (° C) :: $_[0]",
"DEF:mbt1=$log/$dbf.rrd:mbt1:AVERAGE",
"DEF:mbt2=$log/$dbf.rrd:mbt2:AVERAGE",
"LINE1:mbt1$color_mbt1:GPU\\:",
"GPRINT:mbt1:MIN:Minimum\\: % 5.1lf",
"GPRINT:mbt1:MAX:Maximum\\: % 5.1lf",
"GPRINT:mbt1:AVERAGE:Average\\: % 5.1lf",
"GPRINT:mbt1:LAST:Current\\: % 5.1lf °C\\j",
"LINE1:mbt2$color_mbt3:HDD\\:",
"GPRINT:mbt2:MIN:Minimum\\: % 5.1lf",
"GPRINT:mbt2:MAX:Maximum\\: % 5.1lf",
"GPRINT:mbt2:AVERAGE:Average\\: % 5.1lf",
"GPRINT:mbt2:LAST:Current\\: % 5.1lf °C\\j");
$ERROR = RRDs::error;
print "Error: RRDs::graph failed for '$_[0]' : $ERROR\n" if $ERROR;
}
########################################################################
create "$dbf";
gather;
update "$dbf";
graph '2hour', '12hourly';
graph 'day', 'Daily';
graph 'week', 'Weekly';
graph 'month', 'Monthly';
graph 'year', 'Yearly';
########################################################################
# vim: sw=4 ts=8
# End
Oh, and I removed the -v stuff from the last post because it resized the output png making it slightly larger than the others. I also took away some of the graphing options and replaced them with the -A for auto scaling.
_________________
Retired KM user (R4 - R6.04); friend to LH users.
|
|
| Top |
|
|
View unanswered posts
View active topics
Print view
Previous topic Next topic 
LinHES Home
Board index
FAQ
Search
Login 
Register
