Triggered alerts after upgrading to 3.1

[desmobrains]

Hello,

Recently ISE cluster was upgraded to 3.1 from 2.6.

From that time and on, I am receiving the alerts below:

Insufficient Virtual Machine Resources

Details :

On node eacgnvaise03 average IO read performance directly from disk device is: 296 MB/Sec; which is less than the minimum requirement of 300 MB/Sec.  Please update VM hosting to support IO performance requirement.

Description :

Virtual Machine resources such as CPU, RAM, Disk Space, or IOPS are insufficient on this host

Alarm Name :
Insufficient Virtual Machine Resources
Details :
The required minimum of hard disk(s) total size is 300 GB; found only 0 GB on node eacgnvaise04
Description :
Virtual Machine resources such as CPU, RAM, Disk Space, or IOPS are insufficient on this host

Are they valid alerts or we do hit a bug? 

Ideas are welcomed.

Cheers,

Panos

[marce1000]

 - You may confirm the disk test results with :  show tech | begin "disk IO perf"  , for the rest use the monitoring tools from the hypervisor to observe if sufficient resources can remain available for the particular ISE-VM , one might expect some more resource usage on  the new ISE version.

 M.

[desmobrains]

Hi,

I will give it a go and come back.

I just spotted that I get the alert around 03:00 in the morning. I had a look into vCenter and I see spikes on CPU and Disk at that moment...

[marce1000]

  - Presumably  , short single-spike(s) may not need increasing CPU , but if observed several times a  day ,then the conclusion may be opposite.

 M.

[Arne Bier]

The only minimum spec that changed between 2.6 and 3.1 is the minimum RAM for eval - it was 8GB and now it's 16GB.

How much RAM do you have for the VM? And how many CPUs allocated?

The disk IO thing is not an issue unless it's consistently low - 300MB/s is borderline and not that fast to be honest - the more throughput (and more IO) the better for the MNT node. Perhaps ISE is on an oversubscribed host - the ISE resources should not be oversubscribed!

[desmobrains]

Hi Arne,

thanks for coming back.

This is the allocation for each VM:

CPUs:16

RAM: 32GB

HDD: 1,23 TB

The host is not oversubscribed.

Cheers,

Panos 

[Marcelo Morais]

Hi @desmobrains ,

 I agree with @Arne Bier and @marce1000 ...

 ISE continues to test I/O Read and Write Performance on a 3Hr interval (0, 3, 6, 12, 18 and 21 - 9 is skipped due to the CSCvx44981 VM IO Performance Checks not done at 09:00) and ISE has a Daily Task (isedailycron.sh) that takes the Average Disk I/O Performance Measurements gather for the  day and triggers an Alarm if the average is poor than the threshold.

 Your options:

1. Double check if you are receiving the Alarm on ALL Nodes or Nodes of a specific Host

2. Double check the Disk Size - at least 300GB for PSN and 600GB for PAN and MnT (Small or Medium) - take a look at ISE 3.1 Software.

3. open a TAC

Hope this helps !!!

[PetriH8]

Hi  @desmobrains 

Did you find solution for this issue?

[Marcelo Morais]

Hi @PetriH8 ,

 are you having the same issue or something similar ?

[oumodom]

We have the same issue but only specific host with PxGrid, 

Do you think this the bug or anything else?