Monitor Windows via WMI from Cacti on Linux

claymen · Post by **claymen** » Tue Apr 21, 2009 7:36 pm

kosta wrote:Is the graph template out for an 8 core machine yet?

Also, I have been getting these errors since I setup these templates. Some WMI queries work (4 core processors, SQL, memory, context switches, etc, but I get these errors on below templates:

04/20/2009 12:35:14 PM - SPINE: Poller[0] Host[76] ERROR: Empty result [10.x.x.x]: '/usr/local/bin/php -q /var/www/sites/cacti.4rnd.com/branches/cacti-8-7d/scripts/wmi.php 10.X.X.X credential Win32_PerfRawData_PerfProc_Process ThreadCount,HandleCount,PrivateBytes,VirtualBytes,PageFileBytes,PercentPrivilegedTime,PercentProcessorTime,PercentUserTime Name '
04/20/2009 12:35:13 PM - SPINE: Poller[0] Host[76] ERROR: Empty result [10.x.x.x]: '/usr/local/bin/php -q /var/www/sites/cacti.4rnd.com/branches/cacti-8-7d/scripts/wmi.php 10.X.X.X credential Win32_PerfRawData_PerfProc_Process ThreadCount,HandleCount,PrivateBytes,VirtualBytes,PageFileBytes,PercentPrivilegedTime,PercentProcessorTime,PercentUserTime Name '
04/20/2009 12:35:12 PM - SPINE: Poller[0] Host[76] ERROR: Empty result [10.x.x.x]: '/usr/local/bin/php -q /var/www/sites/cacti.4rnd.com/branches/cacti-8-7d/scripts/wmi.php 10.X.X.X credential Win32_PerfRawData_PerfDisk_LogicalDisk DiskWritesPersec,DiskWriteBytesPersec,DiskReadsPersec,DiskReadBytesPersec,CurrentDiskQueueLength Name C'
04/20/2009 12:35:11 PM - SPINE: Poller[0] Host[76] ERROR: Empty result [10.x.x.x]: '/usr/local/bin/php -q /var/www/sites/cacti.4rnd.com/branches/cacti-8-7d/scripts/wmi.php 10.X.X.X credential Win32_LogicalDisk Size,FreeSpace DeviceID C'

Anyone have an idea why?

kosta

I'm guessing you are using pre 0.0.6 and in that case haven't escaped your input correctly. Either upgrade or escape your input. e.g. C: MUST be C\: when put into cacti.

If you upgrade to 0.0.6 this is no longer an issue.

E.g. you'd see something like this

Code: Select all

/usr/bin/php -q /path/to/wmi.php -h 'x.x.x.x' -u '/etc/cacti/cactiwmi.pw' -w 'Win32_LogicalDisk' -n '' -k 'DeviceID' -v 'C:' -c 'Size,FreeSpace'

**Edit**

The 8 core cpu graph is included in 0.0.6

Alternatively you can grab it from here
http://svn.parkingdenied.com/CactiWMI/t ... pu_wmi.xml

But you will need to update the data input method if you wish to continue using the old version. E.g. the templates are suited only for 0.0.6 now however 0.0.6 will work with older templates if you update the credential to be a path to the credential file rather than using wmi-logins.php.

Thomas.Pacce · Post by **Thomas.Pacce** » Wed Apr 22, 2009 7:14 am

Claymen,

Your IIS connections graph shows only Connections.

I only get Connections graphed (only one coloured green graph), shouldnt i get more stuff as the script queries multiple data:
/usr/bin/php -q /usr/share/cacti/site/scripts/wmi.php -h '10.130.16.187' -u '/usr/share/cacti/site/scripts/cactiwmi.pw' -w 'Win32_PerfRawData_W3SVC_WebService' -n '' -k 'Name' -v '_Total' -c 'CurrentConnections,CurrentNonAnonymousUsers,CurrentAnonymousUsers,ConnectionAttemptsPersec,AnonymousUsersPersec,NonAnonymousUsersPersec,LogonAttemptsPersec'
AnonymousUsersPersec:149368 ConnectionAttemptsPersec:19970 CurrentAnonymousUsers:1 CurrentConnections:3 CurrentNonAnonymousUsers:0 LogonAttemptsPersec:149368

I get AnonymousUsersPersec:149368 but thats confusing as the graph shows only CurrentConnections:3

Can you enlight me on this?

russell.seymour · Post by **russell.seymour** » Wed Apr 22, 2009 7:36 am

Hello,

I am getting my system up and running with this addon and things have been going well.

I have come across a discrepancy between the measurement of the CPU Usage. I have my system monitoring with both SNMP and WMI.
The attached files show the graphs that I am seeing.

As you can see the values that each are showing are different. I am inclined to think that the SNMP is the one that is more accurate, but that still begs the question why the WMI graph is playing up.

Additionally it looks like the Total, CPU0 and CPU1 are not in the right order.

Thanks for the great work claymen.
Do you know how cigamit is getting on with the WMI plugin and if your data / graphs are going to be converted to it?

Russell

outspoken · Post by **outspoken** » Wed Apr 22, 2009 10:18 am

Looks like my issue is unresolved. I get 0.00 from 'Total RAM' and 'nan' from everything else in my Memory Usage graph. All other graphs are 'nan' (CPU, Disk I/O, Disk Used, etc).

Target Machine - Windows Enterprise 2003 x64 SP2
Cacti Machine - CentOS 5.2 x86_64
WMI/SNMP installed and running
Cacti 0.8.7d
WMIC Version 4.0.0tp4-SVN-build-UNKNOWN (latest from source)
PHP 5.2.6
Templates 0.0.6
RRDTool 1.2.27

From cacti.log:

Code: Select all

04/22/2009 11:11:11 AM - SPINE: Poller[0] Host[101] DS[1336] WARNING: Result from SCRIPT not valid. Partial Result: ...
04/22/2009 11:11:11 AM - SPINE: Poller[0] Host[101] DS[1336] SCRIPT: /usr/local/bin/php -q /usr/local/www/cacti/scripts/wmi.php -h '10.200.14.26' -u '/etc/cacti/cactiwmi.pw' -w 'Win32_ComputerSystem' -n '' -k '' -v '' -c 'TotalPhysicalMemory', output: U

Running from cli returns:

Code: Select all

Name:SQL09-CV2 TotalPhysicalMemory:17174245376

From wmi debug:

Code: Select all

Time: Wednesday 22nd of April 2009 11:11:37 AM
WMI Class: Win32_ComputerSystem
Credential: /etc/cacti/cactiwmi.pw
Columns: TotalPhysicalMemory
Condition Key:
Condition Val:
Query: "SELECT TotalPhysicalMemory FROM Win32_ComputerSystem"
Exec: /usr/local/bin/wmic --namespace='root\CIMV2' --authentication-file=/etc/cacti/cactiwmi.pw //10.200.14.26 "SELECT TotalPhysicalMemory FROM Win32_ComputerSystem"
Output:
CLASS: Win32_ComputerSystem
Name|TotalPhysicalMemory
SQL09-CV2|17174245376
Output to Cacti: Name:SQL09-CV2 TotalPhysicalMemory:17174245376

RRDTool debug output:

Code: Select all

RRDTool Command:

/usr/local/rrdtool/bin/rrdtool graph - \
--imgformat=PNG \
--start=-86400 \
--end=-300 \
--title="sql09-cv2 - Memory Usage" \
--base=1024 \
--height=120 \
--width=500 \
--alt-autoscale-max \
--lower-limit=0 \
--vertical-label="Bytes" \
--slope-mode \
--font TITLE:8: \
--font AXIS:8: \
--font LEGEND:8: \
--font UNIT:8: \
DEF:a="/usr/local/www/cacti/rra/sql09-cv2_committedbytes_1335.rrd":CommitLimit:AVERAGE \
DEF:b="/usr/local/www/cacti/rra/sql09-cv2_totalphysicalmemory_1336.rrd":TotalPhysicalMemory:AVERAGE \
DEF:c="/usr/local/www/cacti/rra/sql09-cv2_committedbytes_1335.rrd":AvailableBytes:AVERAGE \
DEF:d="/usr/local/www/cacti/rra/sql09-cv2_committedbytes_1335.rrd":CommittedBytes:AVERAGE \
CDEF:cdefi=b,c,- \
AREA:a#562B29FF:"Commit Limit\:"  \
GPRINT:a:LAST:"  Current\:%8.2lf %s"  \
GPRINT:a:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:a:MAX:"Maximum\:%8.2lf %s\n"  \
AREA:b#FF5700FF:"Total RAM\:"  \
GPRINT:b:LAST:"     Current\:%8.2lf %s"  \
GPRINT:b:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:b:MAX:"Maximum\:%8.2lf %s\n"  \
AREA:cdefi#8F005CFF:"Used RAM\:"  \
GPRINT:cdefi:LAST:"      Current\:%8.2lf %s"  \
GPRINT:cdefi:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:cdefi:MAX:"Maximum\:%8.2lf %s\n"  \
LINE2:d#FFFF00FF:"Commited Bytes\:"  \
GPRINT:d:LAST:"Current\:%8.2lf %s"  \
GPRINT:d:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:d:MAX:"Maximum\:%8.2lf %s\n" 

RRDTool Says:

OK

claymen · Post by **claymen** » Wed Apr 22, 2009 8:11 pm

russell.seymour wrote:Hello,

I am getting my system up and running with this addon and things have been going well.

I have come across a discrepancy between the measurement of the CPU Usage. I have my system monitoring with both SNMP and WMI.
The attached files show the graphs that I am seeing.

As you can see the values that each are showing are different. I am inclined to think that the SNMP is the one that is more accurate, but that still begs the question why the WMI graph is playing up.

Additionally it looks like the Total, CPU0 and CPU1 are not in the right order.

Thanks for the great work claymen.
Do you know how cigamit is getting on with the WMI plugin and if your data / graphs are going to be converted to it?

Russell

The WMI class I query for CPU usage is Win32_PerfFormattedData_PerfOS_Processor and the column we use is PercentProcessorTime. This is defined by Microsoft as Percentage of time that the processor is executing a non-idle thread. This property was designed as a primary indicator of processor activity.. You can check it out here http://msdn.microsoft.com/en-us/library ... 85%29.aspx.

We noticed the same thing when we migrated from SNMP to WMI, saw WMI reporting higher CPU usage (more spikey in nature) but overall matched the performance levels we were seeing on the servers. With that said, it should all be taken with a grain of salt considering that your taking a snapshot only every 5 minutes so your averaged graph is never going to be 100% accurate. But with that in mind its long term averages give you an idea of overall performance and allow trending etc.

The total field on the graph is actually supplied directly from WMI. So Cacti isn't actually calculating it from the supplied individual CPU's.

Code: Select all

/usr/bin/php -q /path/wmi-svn.php -h 'x.x.x.x' -u '/etc/cacti/cactiwmi.pw' -w 'Win32_PerfFormattedData_PerfOS_Processor' -n '' -k '' -v '' -c 'PercentProcessorTime' -d1


/usr/local/bin/wmic --namespace='root\CIMV2' --authentication-file=/etc/cacti/cactiwmi.pw //x.x.x.x"SELECT PercentProcessorTime FROM Win32_PerfFormattedData_PerfOS_Processor"
Exec Status: 0

Name0:0
PercentProcessorTime0:24
Name1:1
PercentProcessorTime1:43
Name2:2
PercentProcessorTime2:21
Name3:3
PercentProcessorTime3:48
Name4:_Total

That section above is what WMI returns when querying that as to quite a number of WMI classes. This is why a lot of the time I haven't needed to calculate my own "total" field when WMI provides on which in theory should be always accurate.

Not sure about cigamit's WMI plugin. I did give it a go but it was painful to use and write additional templates for and overall I just found it lacking. It also didn't work without modification and there hasn't been a release since to fix that even though its a known bug. I guess in theory you could port the graphs but you'd still need to sort out the data collection/data templates side of things

claymen · Post by **claymen** » Wed Apr 22, 2009 8:46 pm

Thomas.Pacce wrote:Claymen,

Your IIS connections graph shows only Connections.

I only get Connections graphed (only one coloured green graph), shouldnt i get more stuff as the script queries multiple data:
/usr/bin/php -q /usr/share/cacti/site/scripts/wmi.php -h '10.130.16.187' -u '/usr/share/cacti/site/scripts/cactiwmi.pw' -w 'Win32_PerfRawData_W3SVC_WebService' -n '' -k 'Name' -v '_Total' -c 'CurrentConnections,CurrentNonAnonymousUsers,CurrentAnonymousUsers,ConnectionAttemptsPersec,AnonymousUsersPersec,NonAnonymousUsersPersec,LogonAttemptsPersec'
AnonymousUsersPersec:149368 ConnectionAttemptsPersec:19970 CurrentAnonymousUsers:1 CurrentConnections:3 CurrentNonAnonymousUsers:0 LogonAttemptsPersec:149368

I get AnonymousUsersPersec:149368 but thats confusing as the graph shows only CurrentConnections:3

Can you enlight me on this?

Yes that graph only does connections but we grab other data from the same class because it may be used later on. Your data template can have more data than the graph needs remember. I think I explained it earlier but by grabbing more data in one go you only have the overhead of calling wmic once instead of multiple times (one per column you want to query). The downside is more disk space used but disk space is cheap and time is not when you want your poll time to be as low as possible.

In any case that graph is correct and it only lists connections as it was all that we needed. The data template grabs more which you are welcome to use but you will need to create your own graph

claymen · Post by **claymen** » Wed Apr 22, 2009 8:51 pm

outspoken wrote:Looks like my issue is unresolved. I get 0.00 from 'Total RAM' and 'nan' from everything else in my Memory Usage graph. All other graphs are 'nan' (CPU, Disk I/O, Disk Used, etc).

Target Machine - Windows Enterprise 2003 x64 SP2
Cacti Machine - CentOS 5.2 x86_64
WMI/SNMP installed and running
Cacti 0.8.7d
WMIC Version 4.0.0tp4-SVN-build-UNKNOWN (latest from source)
PHP 5.2.6
Templates 0.0.6
RRDTool 1.2.27

Well the first issue is that cacti isn't pulling the data in so its going to mess the graph. Can you test running the script from the CLI as the user cacti is running as? I can't really see why it's not accepting the output, looks correct to me. The other issue you will have is the CDEF's on that graph will probably break in 0.8.7d which you will need to simply adjust to suit new data inputs cacti gives RRDtool.

outspoken · Post by **outspoken** » Thu Apr 23, 2009 8:58 am

claymen wrote:
outspoken wrote:Looks like my issue is unresolved. I get 0.00 from 'Total RAM' and 'nan' from everything else in my Memory Usage graph. All other graphs are 'nan' (CPU, Disk I/O, Disk Used, etc).

Target Machine - Windows Enterprise 2003 x64 SP2
Cacti Machine - CentOS 5.2 x86_64
WMI/SNMP installed and running
Cacti 0.8.7d
WMIC Version 4.0.0tp4-SVN-build-UNKNOWN (latest from source)
PHP 5.2.6
Templates 0.0.6
RRDTool 1.2.27
Well the first issue is that cacti isn't pulling the data in so its going to mess the graph. Can you test running the script from the CLI as the user cacti is running as? I can't really see why it's not accepting the output, looks correct to me. The other issue you will have is the CDEF's on that graph will probably break in 0.8.7d which you will need to simply adjust to suit new data inputs cacti gives RRDtool.

thanks, this was a permissions issue for the cacti user.

claymen · Post by **claymen** » Thu Apr 23, 2009 9:00 am

outspoken wrote:thanks, this was a permissions issue for the cacti user.

Glad to hear its working

wdennis · Post by **wdennis** » Mon Apr 27, 2009 1:31 pm

Hi all

My environment:
Target Machine - Windows 2003 Server, Std Edn, SP2
Cacti Machine - CentOS release 5.3 (2.6.18-128.1.6.el5)
Cacti 0.8.7d
WMIC Version 4.0.0alpha3-GIT-UNKNOWN
PHP 5.1.6
Templates 0.0.6-r101
RRDTool 1.2.30

I have set up the following graphs based on these templates:
Windows - CPU Usage - 4CPU (WMI)
Windows - Disk I/O (WMI)
Windows - Disk Space (WMI)
Windows - Memory (WMI)
Windows - Total Phyiscal Memory (WMI)

The Disk Space and CPU Usage graphs appear to be working, but the Disk I/O and Memory Usage graphs are not displaying ("red X").

I appear to have wmi.php working correctly:

cacti@cactiserver cacti> /usr/bin/php -q /var/www/cacti/scripts/wmi.php -h 111.222.333.444 -u /etc/cacti/auth.txt -w Win32_PerfRawData_PerfDisk_LogicalDisk -n 'root\CIMV2' -k '' -v '' -c 'DiskWritesPersec,DiskWriteBytesPersec,DiskReadsPersec,DiskReadBytesPersec,CurrentDiskQueueLength'

CurrentDiskQueueLength0:0
DiskReadBytesPersec0:327549485568
DiskReadsPersec0:27095198
DiskWriteBytesPersec0:54619426304
DiskWritesPersec0:3332334
Name0:C
CurrentDiskQueueLength1:0
DiskReadBytesPersec1:327549485568
DiskReadsPersec1:27095198
DiskWriteBytesPersec1:54619

What should I look at?

BTW, it seems that if I turn on the '$dbug = 1;' in wmi.php, the working graphs stop recording data, and there is no log file created in /var/log/cacti/wmi/ which is the $log_location in wmi.php. (The 'cacti' user does have write perm's to that path.) Also, there is no references to wmi.php in the standard cacti log (/var/log/cacti/log)

Any help gratefully appreciated.... This is a wondeful addition to Cacti, and I'm really looking forward to getting this all working!

Thanks,
Will

outspoken · Post by **outspoken** » Mon Apr 27, 2009 2:15 pm

I'm having an issue between versions. This works in 0.8.7d, but not 0.8.7b

working:

Code: Select all

DEF:a="/usr/local/apache/www/cacti2/rra/whcsql19_cpu3_3625.rrd":Total:AVERAGE \
DEF:b="/usr/local/apache/www/cacti2/rra/whcsql19_cpu3_3625.rrd":CPU0:AVERAGE \
DEF:c="/usr/local/apache/www/cacti2/rra/whcsql19_cpu3_3625.rrd":CPU1:AVERAGE \
DEF:d="/usr/local/apache/www/cacti2/rra/whcsql19_cpu3_3625.rrd":CPU2:AVERAGE \
DEF:e="/usr/local/apache/www/cacti2/rra/whcsql19_cpu3_3625.rrd":CPU3:AVERAGE \
AREA:a#FF0000FF:"Total\:"  \
GPRINT:a:LAST:"Current\:%8.2lf %s"  \
GPRINT:a:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:a:MAX:"Maximum\:%8.2lf %s\n"  \
LINE1:b#797C6E7F:"CPU0\:"  \
GPRINT:b:LAST:" Current\:%8.2lf %s"  \
GPRINT:b:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:b:MAX:"Maximum\:%8.2lf %s\n"  \
LINE1:c#0000FF7F:"CPU1\:"  \
GPRINT:c:LAST:" Current\:%8.2lf %s"  \
GPRINT:c:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:c:MAX:"Maximum\:%8.2lf %s\n"  \
LINE1:d#FFAB007F:"CPU2\:"  \
GPRINT:d:LAST:" Current\:%8.2lf %s"  \
GPRINT:d:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:d:MAX:"Maximum\:%8.2lf %s\n"  \
LINE1:e#0000007F:"CPU3\:"  \
GPRINT:e:LAST:" Current\:%8.2lf %s"  \
GPRINT:e:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:e:MAX:"Maximum\:%8.2lf %s\n"

not working:

Code: Select all

DEF:a="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":CPU0:AVERAGE \
DEF:b="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":CPU0:MAX \
DEF:c="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":CPU1:AVERAGE \
DEF:d="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":CPU1:MAX \
DEF:e="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":CPU2:AVERAGE \
DEF:f="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":CPU2:MAX \
DEF:g="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":CPU3:AVERAGE \
DEF:h="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":CPU3:MAX \
DEF:i="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":Total:AVERAGE \
DEF:j="/usr/local/www/cacti/rra/sql01-cv2_cpu1_1463.rrd":Total:MAX \
LINE1:a#797C6EFF:"CPU0\:"  \
GPRINT:a:LAST:" Current\:%8.2lf %s"  \
GPRINT:a:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:b:MAX:"Maximum\:%8.2lf %s\n"  \
LINE1:c#0000FFFF:"CPU1\:"  \
GPRINT:c:LAST:" Current\:%8.2lf %s"  \
GPRINT:c:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:d:MAX:"Maximum\:%8.2lf %s\n"  \
LINE1:e#FFAB00FF:"CPU2\:"  \
GPRINT:e:LAST:" Current\:%8.2lf %s"  \
GPRINT:e:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:f:MAX:"Maximum\:%8.2lf %s\n"  \
LINE1:g#000000FF:"CPU3\:"  \
GPRINT:g:LAST:" Current\:%8.2lf %s"  \
GPRINT:g:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:h:MAX:"Maximum\:%8.2lf %s\n"  \
AREA:i#FF0000B2:"Total\:"  \
GPRINT:i:LAST:"Current\:%8.2lf %s"  \
GPRINT:i:AVERAGE:"Average\:%8.2lf %s"  \
GPRINT:j:MAX:"Maximum\:%8.2lf %s\n"

I noticed that the MAX is a different value from the LAST/AVERAGE on the 'not working' data sources. any idea why this host gives the MAX as a new DEF and how can I remove it, maybe that will fix the problem.

outspoken · Post by **outspoken** » Mon Apr 27, 2009 2:23 pm

I just answered my above question... I went into the Graph Template for CPU and edited the Graph Item Inputs, then removed GPRINT (MAX) from all of the cores.

Not sure if this will make things report incorrectly but the graphs are working now.

brian.nz · Post by **brian.nz** » Mon Apr 27, 2009 9:25 pm

anybody have a compiled version of /usr/local/bin/wmic for solaris sparc < version 10.

stormonts · Post by **stormonts** » Tue Apr 28, 2009 10:51 am

We are running Version 0.8.7b on Linux and MSSQL 2005 on a Windows server.

I was looking for a script / templates so that I could graph the size of each MSSQL database and the total size of a log file with the amount of that file used.

I tried doing this with Erwan's snmptool by assigning OIDs to the various processes, but it wasn't working reliably.

Thanks!

Cacti

Monitor Windows via WMI from Cacti on Linux

Different results between WMI and SNMP for CPU Usage

Re: Different results between WMI and SNMP for CPU Usage

Graphs only kinda working...

Script and or template for MSSQL database size and log util

Who is online