it's pretty hard to find problem without diagnostics, so best way how to solve this is to open case with support.
1. Hardware health is working only for esx hosts, that are managed as node in Orion + you have valid VIM/SAM license
2. how long are you polling data? (recommendations need at least 3 days of data, ideally 1 week and more)
3. do you have VMAN appliance, all data sources added to it, and this appliance integrated with VIM in Orion? with all entities synced via sync wizard?
4. is polling really working? could you take a look into DB, choose one hostid from VIM_Hosts, and then take a look into VIM_HostStatistics_Detail that historical data are really stored in DB?
Peter is on the right track. Please confirm the following:
1. You are polling everything through the VMAN appliance
2. The VMAN appliance is integrated with Orion and synced successfully. You can check synchronization by going to settings - all settings - Virtualization Settings - Run Synchronization Wizard
3. Confirm you have added vcenters instead of the individual hosts as the polling targets.
I am polling through the VMAN appliance.
If you have specific steps on where to click and point I can take a
snapshot or copy screen contents.
The VMAN appliance is integrated with Orion and synched. I have run through
the Run Sync Wizard twice --- last week and again Tuesday.
I did add vCenter(s) and not individual ESXi hosts.
This is a trial install, so I can't, or don't know how, to use Support;
hence the community forum.
I appreciate the assistance. Just can't figure it out, as I followed the
Install guide but just never ended up with live data for "Recommendations"
On Wed, Apr 12, 2017 at 8:15 AM, chrispaap <firstname.lastname@example.org>
This morning the Recommendation Data finally appeared.
Installed on April 4th and results did not show until April 14th. Why it took more than 7 days is unknown.
And the screen itself says data should appear within hours. One actual recommendation was
The CPU utilization on VM ERS-TProd has an average of 7.07% for the last 4 weeks. This is an over-allocated VM which has resources that can be reclaimed.
If it can access 4 weeks of data I'm not sure why it took so long (10 days) to display something.
Still no Hardware Health Data, but that makes sense as I looked into my vCenter and individual hosts and it failed to show any Hardware Sensor Information.
I've been running through every kb article I could find to get that working, though as we have new hardware coming in a few months and these are currently Blades in Chassis all with Alerting in place it isn't a big deal.
I'd suggest the VM team and Orion teams review whether a "Sample" screen for Recommendation displaying for 10 days is the best for end-users.
Once data starts to be reviewed the "sample" data screen be be gone, even if no Recommendations are yet available.
Sample screen says that recommendations could appear in 1 hour, 1 week or 1 month. 1 hour is valid only for the most critical issues, like over-utilized CPU on the host and so on. Non-critical types of recommendations need more data, to create better prediction - we don't want to recommend you to decrease memory for VM and after 1 day to increase it again.
It actually can't access 4 weeks of data - it's just a message, but it's true - i.e. when you have only 1 week of data polled in VIM, average for last 4 weeks is xy% from VIM perspective. "4 weeks" in message should helps users to realize, that in ideal situation we recommend action based on 1 month sample of data.
Still it's interesting, that recommendations showed up after 10 days. Could I ask how big is your environment? (count of clusters, hosts, vms) and how many recommendations were generated?
As you mentioned, hardware health needs sensors on vCenter/host to be working + as I mentioned above, you need valid EVAL/FULL - APM or VIM license in orion, + your ESX Host has to be managed as node in Orion
Hardware Sensors are working now, though as I said, in a blade environment that was not an issue for us.
This is a trial of Virtualization Manager, and thus has a Eval 30-day license.
It is a fairly small environment, 8 Hosts with roughly 75 vms.
It is also a fairly well running environment with only 15 Recommendations, most all related to the number of vCPUs assigned to vms being wasteful.
My point about the sample screen is for the user there is no difference between a sample screen because no data is being processed and the sample screen because the system has not opted to make a recommendation [yet]. This was very annoying for me, and the catalyst for this thread as it passed the 7 day mark and still showed the Sample screen.