In vROps 8 there is a new built-in feature to use instead of Telegraf agents, at least for VMs. You can use the service discovery. This feature, when configured allows vROps to snoop in the VMware tools (subject to version) and discover up to 36 different agents.
VMware tools specific editions are required for this, they are currently:
- 10.3.21 and above for Linux
- 11.0.1 for and above Windows
More information can be found in this KB.
There are a couple of other things that are used depending on the choice of OS for service discovery and Performance Metrics collection:
For Unix: ps, netstat and top for Service Discovery. awk, csh, ps, psgrep and procfs for Performance metrics
Windows: wmic and netstat for Service Discovery and wimic, typeperf and tasklist for Performance Metrics.
Additionally, for non root users, only CPU performance metric is captured/published
On Windows the same counts, the Local Administrator account must be configured. When using Windows AD services, the Domain Admin account must used.
So in other words, know what this requires. It is a nice feature, it makes the overhead easier with agents but some serious rights are required to run this. The other reason to use the fullblown agent is that you do not get as many metrics out of VMTools as you from the telegraf agent. Service Discovery is available in Advanced/Enterprise versions.
Service Discovery is focused on Services as the name suggests, not on what OS is running. It reports back what it finds but focus is on Services.
Configuring this is fairly straight forward. If you go to Home –> Manage Application –> Discover Services you can click on the Configure Service Discovery button. This takes you to the Administration tab and to Cloud Accounts.
Click on the 3 vertical dots to the left of you vCenter server and click edit. Then click on the top most right tab that says Service Discovery. This takes you to the below screen. You can enable Service Discovery by flipping the button by Service Discovery.
Once enabled you can configure how you want to connect to your servers.
Now that this is done, the next step (that you hopefully considered before enabling it is, determining what accounts to use for what.
You can use a default Windows/Linux/SRM Username.
You can use a csv file.
You can use alternate credentials, which is where you setup credentials that you enter and store in vROps like you would when connecting vCenters for example.
So configuring this, I installed two new Linux hosts, just to have some results. While the machines were being discovered by vROps I went and verified that the requested tools were in deed installed (CentOS minimal installation doesn’t always come with all the goodies…). Once that was taken care off I could check the systems in vROps and sure enough aoctest97 and 98 were present. I installed httpd on both servers and enabled the access via the firewall.
aoctest99 was used for testing the Linux Telegraf client so that is why you see it as green already. Now lets configure the credentials for the other two guests and see if we can get some service discovery going.
Entering the credentials and hitting test I get the a successful test. Remember to save the settings at the bottom of the page.
After a little while I did not see anything, so I got curious and sure enough, the version of the open tools is a bit out of date.
Running a yum update changed the version number to:
Going back to the Inventory and waiting to see if something would appear… (It could be a 5 minute interval by default)
(Administration –> Inventory –> Manage Services)
After a little while sure enough, I got some responses, more than I perhaps bargained for.
At first all is unknown, after a little while I got some with common credentials too. Notice that it also detects what is powered on and what is not. Once Service is already discovered, most likely the httpd that I have running.
Lets focus a bit on the test boxes:
All three boxes are from the same template. Only difference was that aoctest97 was upgrade via yum update after the installation. That changed the VMtools to a supported edition and apparently also version from 6 to 7 although the template already is CentOS 7. Repeating the patching of the other two CentOS boxes and a refresh from the adapters they also became visible as CentOS 7. By default Service Monitoring is not enabled. You have to go and enable this. You do it by marking the server(s) and then hit the blue check mark on the menu bar. You then get a green check mark in the Service Monitoring:
Back on the Home tab under Discover Services there is now also a reflection of various services being discovered:
So now that it is installed, what do you use it for.
Now you can quickly locate what services is running on each machine (where you have an ID added). Simply go to the Manage Service, either from the Home tab –> Manage Applications –> Discover Services and click on the service there.
Or go to Administration –> Inventory –> Manage Service and put the search filter in yourself, for example as here, Services.
For now that is all on this topic.