OpenNMS

No, I cannot find much to help you there.
If you have the SC installed,  then SNMP may expose some data to you, but I don't have any data collection configs to send you as starter.
I use vCenter to monitor my virtual infrastructure.

Further to your question however (as it did pique my curiosity):

This looks more like a feature request when you consider the nature of ESX.

ESX(i) exposes some data through SNMP including hardware information it gets via the CIM provider included.
Previous releases that I ran with the service console had limited data available through SNMP (but I only played with that a very tiny bit as vcenter gave me what I needed ...).

Currently the most effective way to get the most indepth access to all ESX metrics is to talk to vcenter through the VMware APIs. VCenter stores all its metrics in a SQL database - including some of the aggregate data it derives from runtime performance counters. VCenter exposes that data through an API (I think there's a java lib for it too).

VCenter nicely handles some of the messy stuff associated with VI performance counter management.

For example:

1.       For VI Admins, CPU ready time is a really important counter, but it's actually derived from other vm-stat-counters by feeding them through a math formula that's tied to the CPU characteristics amongst other variables - it's not something available via snmp AFAIK.

2.       If I perform vMotion on a VM, the network port it's connected to may go to zero (standard vswitch behavior - the vm was unplugged from one port and plugged into another, however there's no link up/down event that correlates with the zeroing of the counters - you need to use the API to correlate the performance counters with ESX / vcenter event logs).

However -- if I'm using distributed vswitches, the network port follows the VM and the counters don't get zeroed

I give these examples as they really illustrates some of the unique aspects of virtual infrastructure monitoring that aren't a concern in a bare-metal/device world.

If you look at other monitoring products, Hyperic for example, it wants you to install an agent on the host that runs the vcenter instance. Using the VMware API, it can fetch and collect all kinds of data from all hypervisors managed by that vcenter instance (max of 1000 esx's per vcenter). I understand that newer vcenter revisions expose the API through a web-interface so agent-less based collection is potentially viable.

To get there however, I think we'd need a new data collection plugin to fetch VMware metrics via the VMware API (Via HTTP or an alternate remote collection protocol). Conceptually it's not impossible, but certainly beyond the scope of this discussion list I think.

You might consider raising a feature request at: http://issues.opennms.org

As I am a contributor and VMWare is something I am very much involved with day-to-day and hence interested in, I may take up the challenge as time permits. But first I'd need a clear understanding of the use cases are for such a collector (ie: what stats? Storage, network, CPU, memory, hardware, virtual machine stats, vswitch stats, etc...), and some of the general functional requirements before I could proceed.

I'd also like to have a discussion on the dev mailing list about the relevance of such a collector:

Anyone using ESX(i)/vCenter knows it can send alert notifications when alarms are fired, only works on hardware it can actually monitor via its built-in CIM providers, and gives you up to 7 years of data retention in an SQL datastore. vSphere client provides the access to the metrics with reporting capabilities and lots of additional third-party device tie-ins via vSphere plugins framework.

Based on that, the use-case for enabling opennms to collect some / all of the vmware performance counters seems limited at best, but I could be wrong there too. If there is a solid need for this type of data collection, are the ESX/vCenter log events also of interest?

Other relevant points of discussion might be:
Would we need to handle the cases where ESX is running virtual appliances (firewalls, load balancers, etc...)?
Would we need to handle the cases where ESX is running non-native components, like the Nexus-1000v dvSwitch replacement?

Some things to consider - I suspect you may not have realized how broad the topic you've touched on really is.

Steve.

From: Kashif Hameed [mailto:kashif.hameed@...]
Sent: Friday, January 28, 2011 7:23 AM
To: opennms-discuss@...
Subject: [opennms-discuss] monitoring esx server

Dear All,

I have installed esx 4.0 server with centos 5.5 server and I have opennms stable veriosn 1.8.7 so is it possible for us to monitor esx server with all virtual machine if yes then plz forward me document that how it is possible

Thanks

hi stephen,

Thanks for your help i will try vcenter for monitoring. By the way my scanrio is that
i have purchased supermicro server with 32 GB ram and 500*3 SATA HD. In it i have installed ESX 4.0 server for virtulization. So now my company has said me to build 20 VM machines on it so have u any advise that what infrastructure i follow in order to build vm machines. Or you can say that can u help me that how could i take more and more performance from virtual machines

Thanks











--- On Fri, 1/28/11, Stephen Hillier <shillier@...> wrote:

From: Stephen Hillier <shillier@...>
Subject: Re: [opennms-discuss] monitoring esx server
To: "General OpenNMS Discussion" <opennms-discuss@...>
Date: Friday, January 28, 2011, 12:06 PM

No, I cannot find much to help you there. If you have the SC installed,  then SNMP may expose some data to you, but I don’t have any data collection configs to send you as starter. I use vCenter to monitor my virtual infrastructure.  Further to your question however (as it did pique my curiosity):   This looks more like a feature request when you consider the nature of ESX.  ESX(i) exposes some data through SNMP including hardware information it gets via the CIM provider included.Previous releases that I ran with the service console had limited data available through SNMP (but I only played with that a very tiny bit as vcenter gave me what I needed …).  Currently the most effective way to get the most indepth access to all ESX metrics is to talk to vcenter through the VMware APIs. VCenter stores all its metrics in a SQL database – including some of the aggregate data it derives from runtime performance counters. VCenter exposes that data
 through an API (I think there’s a java lib for it too).   VCenter nicely handles some of the messy stuff associated with VI performance counter management.   For example:1.       For VI Admins, CPU ready time is a really important counter, but it’s actually derived from other vm-stat-counters by feeding them through a math formula that’s tied to the CPU characteristics amongst other variables – it’s not something available via snmp AFAIK.2.       If I perform vMotion on a VM, the network port it’s connected to may go to zero (standard vswitch behavior – the vm was unplugged from one port and plugged into another, however there’s no link up/down event that correlates with the zeroing of the counters – you need to use the API to correlate the performance counters with ESX / vcenter event logs). However -- if I’m using distributed vswitches, the network port follows the VM and the counters don’t get zeroed  I give these
 examples as they really illustrates some of the unique aspects of virtual infrastructure monitoring that aren’t a concern in a bare-metal/device world.  If you look at other monitoring products, Hyperic for example, it wants you to install an agent on the host that runs the vcenter instance. Using the VMware API, it can fetch and collect all kinds of data from all hypervisors managed by that vcenter instance (max of 1000 esx’s per vcenter). I understand that newer vcenter revisions expose the API through a web-interface so agent-less based collection is potentially viable.  To get there however, I think we’d need a new data collection plugin to fetch VMware metrics via the VMware API (Via HTTP or an alternate remote collection protocol). Conceptually it’s not impossible, but certainly beyond the scope of this discussion list I think.   You might consider raising a feature request at: http://issues.opennms.org  As I am a contributor and
 VMWare is something I am very much involved with day-to-day and hence interested in, I may take up the challenge as time permits. But first I’d need a clear understanding of the use cases are for such a collector (ie: what stats? Storage, network, CPU, memory, hardware, virtual machine stats, vswitch stats, etc…), and some of the general functional requirements before I could proceed.  I’d also like to have a discussion on the dev mailing list about the relevance of such a collector:   Anyone using ESX(i)/vCenter knows it can send alert notifications when alarms are fired, only works on hardware it can actually monitor via its built-in CIM providers, and gives you up to 7 years of data retention in an SQL datastore. vSphere client provides the access to the metrics with reporting capabilities and lots of additional third-party device tie-ins via vSphere plugins framework.  Based on that, the use-case for enabling opennms to collect some / all
 of the vmware performance counters seems limited at best, but I could be wrong there too. If there is a solid need for this type of data collection, are the ESX/vCenter log events also of interest?  Other relevant points of discussion might be: Would we need to handle the cases where ESX is running virtual appliances (firewalls, load balancers, etc…)?Would we need to handle the cases where ESX is running non-native components, like the Nexus-1000v dvSwitch replacement?  Some things to consider – I suspect you may not have realized how broad the topic you’ve touched on really is.  Steve.  From: Kashif Hameed [mailto:kashif.hameed@...] 
Sent: Friday, January 28, 2011 7:23 AM
To: opennms-discuss@...
Subject: [opennms-discuss] monitoring esx server  Dear All,  I have installed esx 4.0 server with centos 5.5 server and I have opennms stable veriosn 1.8.7 so is it possible for us to monitor esx server with all virtual machine if yes then plz forward me document that how it is possible  Thanks  
-----Inline Attachment Follows-----

------------------------------------------------------------------------------
Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)!
Finally, a world-class log management solution at an even better price-free!
Download using promo code Free_Logger_4_Dev2Dev. Offer expires 
February 28th, so secure your free ArcSight Logger TODAY! 
http://p.sf.net/sfu/arcsight-sfd2d
-----Inline Attachment Follows-----

_______________________________________________
Please read the OpenNMS Mailing List FAQ:
http://www.opennms.org/index.php/Mailing_List_FAQ

opennms-discuss mailing list

To *unsubscribe* or change your subscription options, see the bottom of this page:
https://lists.sourceforge.net/lists/listinfo/opennms-discuss

I’m sorry, that’s way off-topic (for this list) and totally beyond the scope of any email I could write. I’m certainly no expert.

If training isn’t an option, then you might want to get a good book or two.
There are some excellent resources to be found online if you know how to `google` for them.
VMware KB is a good starting point, but many VCPs keep blogs with tips and tricks to help you educate yourself.

And a friendly warning: DO NOT virtualize production workloads without some knowledge and at least experience in a lab. It’s not all common sense and it’s really easy to mess things up and then any kind of virtualization gets a bad name because of a poor implementation. Do educate yourself as it’s a non-trivial topic.

Good luck!

Steve.



From: kashif hameed [mailto:kashifhameed_2003@...]
Sent: Saturday, January 29, 2011 12:01 AM
To: General OpenNMS Discussion
Subject: Re: [opennms-discuss] monitoring esx server


hi stephen,

Thanks for your help i will try vcenter for monitoring. By the way my scanrio is that
i have purchased supermicro server with 32 GB ram and 500*3 SATA HD. In it i have installed ESX 4.0 server for virtulization. So now my company has said me to build 20 VM machines on it so have u any advise that what infrastructure i follow in order to build vm machines. Or you can say that can u help me that how could i take more and more performance from virtual machines

Thanks











--- On Fri, 1/28/11, Stephen Hillier <shillier@...> wrote:

From: Stephen Hillier <shillier@...>
Subject: Re: [opennms-discuss] monitoring esx server
To: "General OpenNMS Discussion" <opennms-discuss@...>
Date: Friday, January 28, 2011, 12:06 PM

No, I cannot find much to help you there.

If you have the SC installed,  then SNMP may expose some data to you, but I don’t have any data collection configs to send you as starter.

I use vCenter to monitor my virtual infrastructure.



Further to your question however (as it did pique my curiosity):



This looks more like a feature request when you consider the nature of ESX.



ESX(i) exposes some data through SNMP including hardware information it gets via the CIM provider included.

Previous releases that I ran with the service console had limited data available through SNMP (but I only played with that a very tiny bit as vcenter gave me what I needed …).



Currently the most effective way to get the most indepth access to all ESX metrics is to talk to vcenter through the VMware APIs. VCenter stores all its metrics in a SQL database – including some of the aggregate data it derives from runtime performance counters. VCenter exposes that data through an API (I think there’s a java lib for it too).



VCenter nicely handles some of the messy stuff associated with VI performance counter management.



For example:

1.       For VI Admins, CPU ready time is a really important counter, but it’s actually derived from other vm-stat-counters by feeding them through a math formula that’s tied to the CPU characteristics amongst other variables – it’s not something available via snmp AFAIK.

2.       If I perform vMotion on a VM, the network port it’s connected to may go to zero (standard vswitch behavior – the vm was unplugged from one port and plugged into another, however there’s no link up/down event that correlates with the zeroing of the counters – you need to use the API to correlate the performance counters with ESX / vcenter event logs).

However -- if I’m using distributed vswitches, the network port follows the VM and the counters don’t get zeroed



I give these examples as they really illustrates some of the unique aspects of virtual infrastructure monitoring that aren’t a concern in a bare-metal/device world.



If you look at other monitoring products, Hyperic for example, it wants you to install an agent on the host that runs the vcenter instance. Using the VMware API, it can fetch and collect all kinds of data from all hypervisors managed by that vcenter instance (max of 1000 esx’s per vcenter). I understand that newer vcenter revisions expose the API through a web-interface so agent-less based collection is potentially viable.



To get there however, I think we’d need a new data collection plugin to fetch VMware metrics via the VMware API (Via HTTP or an alternate remote collection protocol). Conceptually it’s not impossible, but certainly beyond the scope of this discussion list I think.



You might consider raising a feature request at: http://issues.opennms.org



As I am a contributor and VMWare is something I am very much involved with day-to-day and hence interested in, I may take up the challenge as time permits. But first I’d need a clear understanding of the use cases are for such a collector (ie: what stats? Storage, network, CPU, memory, hardware, virtual machine stats, vswitch stats, etc…), and some of the general functional requirements before I could proceed.



I’d also like to have a discussion on the dev mailing list about the relevance of such a collector:



Anyone using ESX(i)/vCenter knows it can send alert notifications when alarms are fired, only works on hardware it can actually monitor via its built-in CIM providers, and gives you up to 7 years of data retention in an SQL datastore. vSphere client provides the access to the metrics with reporting capabilities and lots of additional third-party device tie-ins via vSphere plugins framework.



Based on that, the use-case for enabling opennms to collect some / all of the vmware performance counters seems limited at best, but I could be wrong there too. If there is a solid need for this type of data collection, are the ESX/vCenter log events also of interest?



Other relevant points of discussion might be:

Would we need to handle the cases where ESX is running virtual appliances (firewalls, load balancers, etc…)?

Would we need to handle the cases where ESX is running non-native components, like the Nexus-1000v dvSwitch replacement?



Some things to consider – I suspect you may not have realized how broad the topic you’ve touched on really is.



Steve.



From: Kashif Hameed [mailto:kashif.hameed@...]
Sent: Friday, January 28, 2011 7:23 AM
To: opennms-discuss@...
Subject: [opennms-discuss] monitoring esx server



Dear All,



I have installed esx 4.0 server with centos 5.5 server and I have opennms stable veriosn 1.8.7 so is it possible for us to monitor esx server with all virtual machine if yes then plz forward me document that how it is possible



Thanks



-----Inline Attachment Follows-----
------------------------------------------------------------------------------
Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)!
Finally, a world-class log management solution at an even better price-free!
Download using promo code Free_Logger_4_Dev2Dev. Offer expires
February 28th, so secure your free ArcSight Logger TODAY!
http://p.sf.net/sfu/arcsight-sfd2d

-----Inline Attachment Follows-----
_______________________________________________
Please read the OpenNMS Mailing List FAQ:
http://www.opennms.org/index.php/Mailing_List_FAQ

opennms-discuss mailing list

To *unsubscribe* or change your subscription options, see the bottom of this page:
https://lists.sourceforge.net/lists/listinfo/opennms-discuss

ok thanks


--- On Fri, 1/28/11, Stephen Hillier <shillier@...> wrote:

From: Stephen Hillier <shillier@...>
Subject: Re: [opennms-discuss] monitoring esx server
To: "General OpenNMS Discussion" <opennms-discuss@...>
Date: Friday, January 28, 2011, 10:44 PM

I’m sorry, that’s way off-topic (for this list) and totally beyond the scope of any email I could write. I’m certainly no expert.  If training isn’t an option, then you might want to get a good book or two. There are some excellent resources to be found online if you know how to `google` for them. VMware KB is a good starting point, but many VCPs keep blogs with tips and tricks to help you educate yourself.  And a friendly warning: DO NOT virtualize production workloads without some knowledge and at least experience in a lab. It’s not all common sense and it’s really easy to mess things up and then any kind of virtualization gets a bad name because of a poor implementation. Do educate yourself as it’s a non-trivial topic.  Good luck!  Steve.      From: kashif hameed [mailto:kashifhameed_2003@...] 
Sent: Saturday, January 29, 2011 12:01 AM
To: General OpenNMS Discussion
Subject: Re: [opennms-discuss] monitoring esx server  
hi stephen,

Thanks for your help i will try vcenter for monitoring. By the way my scanrio is that
i have purchased supermicro server with 32 GB ram and 500*3 SATA HD. In it i have installed ESX 4.0 server for virtulization. So now my company has said me to build 20 VM machines on it so have u any advise that what infrastructure i follow in order to build vm machines. Or you can say that can u help me that how could i take more and more performance from virtual machines

Thanks











--- On Fri, 1/28/11, Stephen Hillier <shillier@...> wrote:
From: Stephen Hillier <shillier@...>
Subject: Re: [opennms-discuss] monitoring esx server
To: "General OpenNMS Discussion" <opennms-discuss@...>
Date: Friday, January 28, 2011, 12:06 PMNo, I cannot find much to help you there. If you have the SC installed,  then SNMP may expose some data to you, but I don’t have any data collection configs to send you as starter. I use vCenter to monitor my virtual infrastructure. Further to your question however (as it did pique my curiosity):  This looks more like a feature request when you consider the nature of ESX. ESX(i) exposes some data through SNMP including hardware information it gets via the CIM provider included.Previous releases that I ran with the service console had limited data available through SNMP (but I only played with that a very tiny bit as vcenter gave me what I needed …). Currently the most effective way to get the most indepth access to all ESX metrics is to talk to vcenter through the VMware APIs. VCenter stores all its metrics in a SQL database – including some of the aggregate data it derives from runtime performance
 counters. VCenter exposes that data through an API (I think there’s a java lib for it too).  VCenter nicely handles some of the messy stuff associated with VI performance counter management.  For example:1.       For VI Admins, CPU ready time is a really important counter, but it’s actually derived from other vm-stat-counters by feeding them through a math formula that’s tied to the CPU characteristics amongst other variables – it’s not something available via snmp AFAIK.2.       If I perform vMotion on a VM, the network port it’s connected to may go to zero (standard vswitch behavior – the vm was unplugged from one port and plugged into another, however there’s no link up/down event that correlates with the zeroing of the counters – you need to use the API to correlate the performance counters with ESX / vcenter event logs). However -- if I’m using distributed vswitches, the network port follows the VM and the counters
 don’t get zeroed I give these examples as they really illustrates some of the unique aspects of virtual infrastructure monitoring that aren’t a concern in a bare-metal/device world. If you look at other monitoring products, Hyperic for example, it wants you to install an agent on the host that runs the vcenter instance. Using the VMware API, it can fetch and collect all kinds of data from all hypervisors managed by that vcenter instance (max of 1000 esx’s per vcenter). I understand that newer vcenter revisions expose the API through a web-interface so agent-less based collection is potentially viable. To get there however, I think we’d need a new data collection plugin to fetch VMware metrics via the VMware API (Via HTTP or an alternate remote collection protocol). Conceptually it’s not impossible, but certainly beyond the scope of this discussion list I think.  You might consider raising a feature request at: http://issues.opennms.org As
 I am a contributor and VMWare is something I am very much involved with day-to-day and hence interested in, I may take up the challenge as time permits. But first I’d need a clear understanding of the use cases are for such a collector (ie: what stats? Storage, network, CPU, memory, hardware, virtual machine stats, vswitch stats, etc…), and some of the general functional requirements before I could proceed. I’d also like to have a discussion on the dev mailing list about the relevance of such a collector:  Anyone using ESX(i)/vCenter knows it can send alert notifications when alarms are fired, only works on hardware it can actually monitor via its built-in CIM providers, and gives you up to 7 years of data retention in an SQL datastore. vSphere client provides the access to the metrics with reporting capabilities and lots of additional third-party device tie-ins via vSphere plugins framework. Based on that, the use-case for enabling opennms to
 collect some / all of the vmware performance counters seems limited at best, but I could be wrong there too. If there is a solid need for this type of data collection, are the ESX/vCenter log events also of interest? Other relevant points of discussion might be: Would we need to handle the cases where ESX is running virtual appliances (firewalls, load balancers, etc…)?Would we need to handle the cases where ESX is running non-native components, like the Nexus-1000v dvSwitch replacement? Some things to consider – I suspect you may not have realized how broad the topic you’ve touched on really is. Steve. From: Kashif Hameed [mailto:kashif.hameed@...] 
Sent: Friday, January 28, 2011 7:23 AM
To: opennms-discuss@...
Subject: [opennms-discuss] monitoring esx server Dear All, I have installed esx 4.0 server with centos 5.5 server and I have opennms stable veriosn 1.8.7 so is it possible for us to monitor esx server with all virtual machine if yes then plz forward me document that how it is possible Thanks 
-----Inline Attachment Follows-----------------------------------------------------------------------------------
Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)!
Finally, a world-class log management solution at an even better price-free!
Download using promo code Free_Logger_4_Dev2Dev. Offer expires 
February 28th, so secure your free ArcSight Logger TODAY! 
http://p.sf.net/sfu/arcsight-sfd2d
-----Inline Attachment Follows-----_______________________________________________
Please read the OpenNMS Mailing List FAQ:
http://www.opennms.org/index.php/Mailing_List_FAQ

opennms-discuss mailing list

To *unsubscribe* or change your subscription options, see the bottom of this page:
https://lists.sourceforge.net/lists/listinfo/opennms-discuss  
-----Inline Attachment Follows-----

------------------------------------------------------------------------------
Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)!
Finally, a world-class log management solution at an even better price-free!
Download using promo code Free_Logger_4_Dev2Dev. Offer expires 
February 28th, so secure your free ArcSight Logger TODAY! 
http://p.sf.net/sfu/arcsight-sfd2d
-----Inline Attachment Follows-----

_______________________________________________
Please read the OpenNMS Mailing List FAQ:
http://www.opennms.org/index.php/Mailing_List_FAQ

opennms-discuss mailing list

To *unsubscribe* or change your subscription options, see the bottom of this page:
https://lists.sourceforge.net/lists/listinfo/opennms-discuss