ESX 3.5 HA Problem

[RKuehn]

After updateing from VC2.02 to VC2.5 and ESX3.02 to ESX3.5 I got an error message from HA.

"Configuration Issues

Insufficent resources to satisfy HA failover level on cluster "Cluster"

host xy currently has no management network redundancy"

Why is this happen?

I checked the Resource Allocation and almost all vm's are configured as "Normal" Shares.

When I changed the "Shares" from "Custom" to "Normal" the "Shares Value" is changing from 1024 to 655360.

Is this normal?

What can I do, to solve the Configuration Issues?

Regards

Rainer

[FredPeterson]

Does your VMKernel vSwitch have multiple network adapters assigned for failover?

[RKuehn]

I would prefer to use only one Network card.

Is it possible to turn off this message?

The resource problem still exists.

If I configured the HA-Function I can't start any VM.

"Insufficient resources to satisfy configured failover level for HA"

But there is enough Mem/CPU Resources available.

[man_gold]

even configuring two adapters for the service console does not change anything.

@RKUEHN: the only workaround is to edit the HA-Cluster settings to "Allow virtual machines to be powered on even if they violate aviability constrains"

[admin]

Did you try reconfiguring HA on the host on which you added the extra nic? You'll need to do that to make the warning go away.

[admin]

What is the maximum cpu and memory reservation among all powered on vms in the cluster?

[man_gold]

yes you're right, after reconfiguring for HA the warning disappeared. Thanks

[sg222]

Same problem here.

Situation:

we have two ESX Servers. Both with 64gb ram and 16 cores.

Both esx hosts are part of a vmware cluster with ha and drs activated.

We have three resource-pools.

1. production

-- cpu shares high

-- mem shares high

no reservation

2. test and development

-- cpu shares low

-- mem shares low

-- 4.096mb ram reserved

3. administration

-- cpu shares high

-- mem shares high

-- 1.024mb reserved ram

Host a hosts the following virtual machines in resource pool production (HA and DRS disabled for these machines):

- VM A, 16.384mb configured ram, 8.192 mb reserved ram

- VM B, 16.384mb configured ram, 8.192 mb reserved ram

- VM C, 16.384mb configured ram, 8.192 mb reserved ram

in resource pool development:

- VM D with 4.096mb configured ram, no reservation, HA and DRS fully automated

VM E, placed in development resource pool, ha and drs activated does not start!!! I thought the ESX would automatically start this VM on the other host, which only hosts 2 small machines.

A popup appears "insufficient resources to satisfy configured failover level for ha".

So what does the HA-option "Allow virtual machines to be powered on even if they violate availability constraints" actually mean? To allow power on virtual machines if no physical ram is left? But we have physical ram left, even if you count configured ram:

running machines:

- 16.384 mb ram * 3 = 49.152 mb ram (3 big VMs)

+ 4.096 mb configured ram = 53.248 mb ram in use on esx host A. So also host a would be able to start VM E - but it does not, it even does not try it on esx host B?!

[Amnexi]

Reached here looking for other stuff. I'm surely posting too late, but

here I go: I think you had an atypical Shares/Reservation/Resource Pool

configuration that made HA calculations result in "insufficient resources" value.

I'm not sure, but maybe part of the problem would be:

1) 2 VMs with Reservation defined, within a RP with no reservation (Production).

2) 4GB RAM reserved in T&D RP, and the only VM within that RP is configured with all that amount of RAM.

3) Also, remember that HA constraints are calculated comparing the most

RAM/CPU consuming VM with the less RAM/CPU-free ESX host, and you have

big VMs...

I

suggest following Best Practices described in Chapter 10 of the

Resource Management Guide

(http://www.vmware.com/pdf/vi3_301_201_resource_mgmt.pdf), for example

related with point 2) described above:

Don’t set Reservation too high. A reservation that’s too high can limit the number

of virtual machines in a resource pool

I hope this helps a little bit...

[bluesland]

I had the same problem. I setup redundant nic's and vswitches for service console, VMotion and Virtual machines connected to separate physical switches. I increased the service console memory to 800MB and the swap to 1600MB. I also reconfigured the service console partitions and sizes for each host to prevent the root partition getting filled up. I isolated my /boot, / and /swap from all other partitions. There are many suggestion out there.

[Amnexi]

You're right, lots of sutff here. I was actually replying only to sg222. Same redundant config here: 2nic's for SC, 1independent for VMotion, 4 connected to different pSwitches for VMs, 800MB SC, increased partition sizes.

Regards

[Erik_Zandboer]

Hi,

Using dual NICs for the SC should indeed solve the problem when HA states there is no redundancy. You SHOULD always use two NICs, over two separate IP switches. If you plan to have HA enabled, you've got to have the underlying hardware as well (redundant that is).

If you have questions on how HA calculates failover capacity, I wrote a blog post on that:

Visit my blog at