Win2k / Win2k3 Multi CPU / Hyperthreading CPU Graphing ?
Moderators: Developers, Moderators
Win2k / Win2k3 Multi CPU / Hyperthreading CPU Graphing ?
Currently our company uses perfmon to watch CPU load across all windows servers, we'd like to replace it with Cacti/RRDTool
But we don't like that we must graph our HT dual CPU compaq DL360's and higher boxes as 4 different graphs (2 HT CPUs)
4 graphs doesnt really tell us the real 'cpu load' of a server.
how can you obtain this info in one graph?
I've done searching on this site and havent found a 'definitive' answer to this problem.
But we don't like that we must graph our HT dual CPU compaq DL360's and higher boxes as 4 different graphs (2 HT CPUs)
4 graphs doesnt really tell us the real 'cpu load' of a server.
how can you obtain this info in one graph?
I've done searching on this site and havent found a 'definitive' answer to this problem.
- TheWitness
- Developer
- Posts: 17061
- Joined: Tue May 14, 2002 5:08 pm
- Location: MI, USA
- Contact:
Ditto
True understanding begins only when we realize how little we truly understand...
Life is an adventure, let yours begin with Cacti!
Author of dozens of Cacti plugins and customization's. Advocate of LAMP, MariaDB, IBM Spectrum LSF and the world of batch. Creator of IBM Spectrum RTM, author of quite a bit of unpublished work and most of Cacti's bugs.
_________________
Official Cacti Documentation
GitHub Repository with Supported Plugins
Percona Device Packages (no support)
Interesting Device Packages
For those wondering, I'm still here, but lost in the shadows. Yearning for less bugs. Who want's a Cacti 1.3/2.0? Streams anyone?
Life is an adventure, let yours begin with Cacti!
Author of dozens of Cacti plugins and customization's. Advocate of LAMP, MariaDB, IBM Spectrum LSF and the world of batch. Creator of IBM Spectrum RTM, author of quite a bit of unpublished work and most of Cacti's bugs.
_________________
Official Cacti Documentation
GitHub Repository with Supported Plugins
Percona Device Packages (no support)
Interesting Device Packages
For those wondering, I'm still here, but lost in the shadows. Yearning for less bugs. Who want's a Cacti 1.3/2.0? Streams anyone?
Details
Sorry about not being clear.
Basically I want a single graph that displays percent CPU load for the entire system like I can get from a Unix Box.
From a windows host, I am only able to get one graph per CPU, and in the case HT, I get 2 graphs per CPU, which makes it very hard to tell the actual 'CPU Load' of a windows server at a glance.
I'd be happy with either of the following as a solution:
1. Ideally, a total CPU load that looks identical to the ones I get from Unix boxes.
2. Or at least one graph that charts all the loads from each CPU one on graph (each cpu being a line), rather then be spread acrosss 2-6 graphs.
PS: Witness, get on AIM
Basically I want a single graph that displays percent CPU load for the entire system like I can get from a Unix Box.
From a windows host, I am only able to get one graph per CPU, and in the case HT, I get 2 graphs per CPU, which makes it very hard to tell the actual 'CPU Load' of a windows server at a glance.
I'd be happy with either of the following as a solution:
1. Ideally, a total CPU load that looks identical to the ones I get from Unix boxes.
2. Or at least one graph that charts all the loads from each CPU one on graph (each cpu being a line), rather then be spread acrosss 2-6 graphs.
PS: Witness, get on AIM
Here ya go, if you can use WMI
Doh..
that work is awesome of yours DevilSun, but due the limitations of this environment, I really need to find a SNMP only solution if one exists...
since I am running cacti on linux.
since I am running cacti on linux.
- TheWitness
- Developer
- Posts: 17061
- Joined: Tue May 14, 2002 5:08 pm
- Location: MI, USA
- Contact:
What I have done to accomodate this is once I have created the individual DS's, I create a new graph based upon their contents. There is another approach to create a Window/XP Host template for 1, 2, 4 and 8 way systems and to create a graph template for that host that includes all CPU's.
I have done it manually to date and it works GREEAAATT on both our 4 way and 8 way boxes.
TheWitness
I have done it manually to date and it works GREEAAATT on both our 4 way and 8 way boxes.
TheWitness
True understanding begins only when we realize how little we truly understand...
Life is an adventure, let yours begin with Cacti!
Author of dozens of Cacti plugins and customization's. Advocate of LAMP, MariaDB, IBM Spectrum LSF and the world of batch. Creator of IBM Spectrum RTM, author of quite a bit of unpublished work and most of Cacti's bugs.
_________________
Official Cacti Documentation
GitHub Repository with Supported Plugins
Percona Device Packages (no support)
Interesting Device Packages
For those wondering, I'm still here, but lost in the shadows. Yearning for less bugs. Who want's a Cacti 1.3/2.0? Streams anyone?
Life is an adventure, let yours begin with Cacti!
Author of dozens of Cacti plugins and customization's. Advocate of LAMP, MariaDB, IBM Spectrum LSF and the world of batch. Creator of IBM Spectrum RTM, author of quite a bit of unpublished work and most of Cacti's bugs.
_________________
Official Cacti Documentation
GitHub Repository with Supported Plugins
Percona Device Packages (no support)
Interesting Device Packages
For those wondering, I'm still here, but lost in the shadows. Yearning for less bugs. Who want's a Cacti 1.3/2.0? Streams anyone?
Not exactly what you may be after but you could try - http://forums.cacti.net/viewtopic.php?t=4668
One other option is to load another agent on the servers. We're currently using one from that I found at http://www.snmp-informant.com/
It allows you to poll CPU, Disk, Memory, and Network via SNMP and it's free. Works well for us. They also have an advanced agent that reports on most everything that perfmon does( it's not free but relatively cheap)
The other option for you would be to look at HP's MIB's and see what you can get from there. They may have a total CPU load as well and if you have the insight agents loaded, then they are already displaying that data.
BTW, the agent from snmpinformant seems pretty stable. I'm currently running it on about 30 Windows 2000/2003 servers with no problems at all. Planning on rolling it out to about 500+ in the next couple of months using Cacti to do the perf graphing...
Good luck!
It allows you to poll CPU, Disk, Memory, and Network via SNMP and it's free. Works well for us. They also have an advanced agent that reports on most everything that perfmon does( it's not free but relatively cheap)
The other option for you would be to look at HP's MIB's and see what you can get from there. They may have a total CPU load as well and if you have the insight agents loaded, then they are already displaying that data.
BTW, the agent from snmpinformant seems pretty stable. I'm currently running it on about 30 Windows 2000/2003 servers with no problems at all. Planning on rolling it out to about 500+ in the next couple of months using Cacti to do the perf graphing...
Good luck!
Windows Data Template
Skaminski,
Could you post your data / graph / host templates for your SNMP-INFORMANT queries?
Could you post your data / graph / host templates for your SNMP-INFORMANT queries?
I don't think my templates will work unless you purchase the advanced agent from them. However, i can post some of my graphs when i get in the office. I don't have a problem putting anything up here to help. If you'd like the templates for the advanced agent, I'll put them up though.
I could also help you with the one for the free agent and then you could post it up. Configuring it is pretty simple.
Sean
I could also help you with the one for the free agent and then you could post it up. Configuring it is pretty simple.
Sean
I don't have any using the free SNMP Informant agent. Setting everything up is fairly easy though. Download and install the agent on the machine you want to monitor. Copy the MIB's from SNMP informant into your MIB folder of Getif (or whatever you use) delete the .index file and restart getif.
From within Getif, you should now be able to browse the OID Tree and will see the metrics provided via the agent. I'll post the templates for CPU Total %Utilization that I'm using. However, if you use the free version, I belive they have a differnt OID so you may have to modify the OID string so that it will work correctly for you.
From within Getif, you should now be able to browse the OID Tree and will see the metrics provided via the agent. I'll post the templates for CPU Total %Utilization that I'm using. However, if you use the free version, I belive they have a differnt OID so you may have to modify the OID string so that it will work correctly for you.
- Attachments
-
- cacti_data_template_windows_cpu_total_utilization.xml
- Windows CPU % Utilization Data Template
- (5.56 KiB) Downloaded 521 times
-
- cacti_graph_template_windows_cpu_total_utilization.xml
- Windows CPU% Utilization Graph Template
- (10.47 KiB) Downloaded 625 times
-
- Picture of Graph. I'll soon be adding some other counters to this that give some more data regarding CPU Utilizatiion and Load
- graph_image.php.png (4.09 KiB) Viewed 8824 times
After reading your post, I saw you were also looking for Disk, Memory, and Network.
For Disk and Network, I use the standard Queries SNMP - Interface Statistics and SNMP- Get Mounted Partitions. They seem to work well and since I have different servers with different NIC's and Disk configurations, these queries give me the data I need.
For Memory, I've created one similar to the CPU graph. I'll also be adding some additional memory counters related to paging and memory pools (paged and non paged)
For Disk and Network, I use the standard Queries SNMP - Interface Statistics and SNMP- Get Mounted Partitions. They seem to work well and since I have different servers with different NIC's and Disk configurations, these queries give me the data I need.
For Memory, I've created one similar to the CPU graph. I'll also be adding some additional memory counters related to paging and memory pools (paged and non paged)
http://forums.cacti.net/viewtopic.php?t=4668 uses the free version of snmp informant for
% DPC Time
% Interrupt Time
% Privileged Time
% Processor Time
% User Time
per processor and one for total
% DPC Time
% Interrupt Time
% Privileged Time
% Processor Time
% User Time
per processor and one for total
Who is online
Users browsing this forum: No registered users and 2 guests