RHEV cpu and memory reporting issue


#1

Hi,

I can’t get any data in MIQ about CPU (CPU - Usage Rate (%) (Avg)) and memory (Memory - Avg Used for Collected Intervals (MB) (Avg)) on RHEV clusters. C/U is configured. Is it a known issue ? Do i need to configure something special to get these data in MIQ ?
I can see data about cpu and memory with the RHEV engine report.

Thanks !

Michael


Utilizations graphs for VMs on RHV are empty
#2

Are you seeing any errors in the logs?


#3

I don’t see any errors in the logs but as i’m new to MIQ, i might missing something.

In configure - configuration - diagnostic, under worker, i can see that the rhev collector as collected information about cpu and memory.


#4

Hi mlessard,

You mentioned you configured C&U, but did you add the C&U database credentials into ManageIQ?

You can learn more about setting that up here:
https://access.redhat.com/documentation/en-US/CloudForms/3.0/html-single/Management_Engine_5.2_Insight_Guide/index.html#Notes_on_Capacity_and_Utilization_Collection_for_RHEVM_3.1
https://access.redhat.com/solutions/63277

-Alex


#5

Yes, i did it. When i click on validate, everything is good.

Is there a way to validate that MIQ can read cpu and memory data from RHEV other then trying to generate reports ?


#6

@mlessard You can validate by navigating to a virtual machine under (Infrastructure -> Virtual Machines) then selecting Monitoring -> Utilization. Depending on how long MIQ has been connected, you may have to change the interval to Hourly or most recent Hour to see collected data.

Here’s the documentation:
https://access.redhat.com/documentation/en-US/CloudForms/3.0/html-single/Management_Engine_5.2_Insight_Guide/index.html#To_view_capacity_and_utilization_charts_for_a_Virtual_Machine

-Alex


#7

Alex, i can’t select Utilization, it’s in gray.


#8

@mlessard Check that C&U collections was turned on for the region.

Navigate to Config -> Configuration on the top navigation tabs. Select Region in the left side accordion and look for C&U collection in the displayed panel. Check the boxes for “Collect for All Clusters”/“Collect for All Datastores” or check the clusters/datastores you desire.

-Alex


#9

Yes, they were already selected.


#10

Are there any errors in evm.log? Also grep for the text “no metrics” in evm.log to see if the perf_capture messages are not retrieving any data.


#11

Yes, there is a bunch of no metrics message :

INFO Realtime skipping processing for host Redhat name : rhevh1 id 111212121212 as no metrics were captured.


#12

Check that dwhd (Data warehousing damon is running on your rhevm machine)

# service ovirt-engine-dwhd status
/etc/init.d/ovirt-engine-dwhd is running (pid 3798)

#13

Yes it runs :
[root@rhevmanager ~]# service ovirt-engine-dwhd status
ovirt-engine-dwhd (pid 31209) is running…


#14

In this document: https://access.redhat.com/documentation/en-US/CloudForms/3.0/html/Management_Engine_5.2_Release_Notes/chap-Support_Matrix.html

There is lot of items where RHEV is not supported, could it be related to some of those items ? Like Alerts - Real Time by example.


#15

C&U is supported for RHEVM environments. I’m curious if your environment is actually collecting the samples and placing them in the ovirt_engine_history database on your rhevm machine. You can check this on the rhevm machine by querying the postgres database for samples of a specific virtual machine:

# su postgres
bash-4.1$ psql -d ovirt_engine_history
psql (8.4.20)
Type "help" for help.
ovirt_engine_history=# SELECT a.history_datetime, a.cpu_usage_percent, a.memory_usage_percent FROM vm_samples_history AS a, vm_configuration AS b WHERE a.vm_id = b.vm_id AND b.vm_name = 'VM-NAME' ORDER BY a.history_datetime DESC limit 10;

      history_datetime      | cpu_usage_percent | memory_usage_percent
----------------------------+-------------------+----------------------
 2014-09-22 20:35:27.207-04 |                 0 |                   34
 2014-09-22 20:34:27.205-04 |                 0 |                   34
 2014-09-22 20:33:27.206-04 |                 0 |                   34
 2014-09-22 20:32:27.205-04 |                 0 |                   33
 2014-09-22 20:31:27.203-04 |                 0 |                   34
 2014-09-22 20:30:27.203-04 |                 1 |                   33
 2014-09-22 20:29:27.202-04 |                 1 |                   33
 2014-09-22 20:28:27.2-04   |                 0 |                   33
 2014-09-22 20:27:27.2-04   |                 0 |                   33
 2014-09-22 20:26:27.197-04 |                 0 |                   34
(10 rows)

In the above example replace VM-NAME with the VM your trying to view CPU/Memory C&U data on.

I also noticed for the name of your host you have rhevh, are your host machines Red Hat Enterprise Virtualization Hypervisors or Red Hat Enterprise Linux Hosts? I would have to find out if ovirt_engine_history samples are collected in environments with Hypervisors vs actual full Red Hat Enterprise Linux Hosts.

-Alex


#16

I get this :

history_datetime | cpu_usage_percent | memory_usage_percent
------------------±------------------±---------------------
(0 rows)

I have two Red hat Enetreprise Linux Hosts (i’m not using rhev-h really, it’s just names).


#17

I have the same problem.

Every time i run the widget report “Top CPU consumers” and “Top Memory Consumers” the result are the same: “No records found”

I’m recollecting C&U data without problem from all clusters, hosts and vm’s.

Any idea?

Thanks


#18

@gstaricco

Have you confirmed C&U data is on the appliance before running the widget report?

In mlessard’s case an upgrade to RHEVM had inadvertently changed the dwhd’s database name so that ManageIQ would not find the metrics. This is likely a rare case.


#19

Yes, the C&U is on the appliance before the running of the report.

Maybe is the version of de MIQ (pre-build appliance). Last week RedHat installed in my enterprise a trial of Cloudforms and the reports are working fine.

I will try to install the last development version of MIQ and check if the reports works in the last version.

::::::::::::

Well i dont know how, but i delete the defaults reports from the manageiqweb, and i created them again with the same options and now are working fine.


#20

I am having the same problem, I can see all the VMs but it doesn’t get any data, I just downloaded the last apliance version and I’m going to try it again, I’ll let you know if works.