VMM 2.1 causing VM crash

[Ole Thomsen_1]

Thanks Rob,

I installed VMM 3.0 two days ago.

We have a small VCB job running every night with 3 servers, and the first job completed without errors.

But tonight a server was powered down, after these events:

Cannot open the disk /vmfs/...... or one of the snapshot disks it depends on. Reason: device is busy

Failed to reopen disk (null)

Error encountered while restarting virtual machine after taking snapshot. The virtual machine will be powered off.

Doing an online commit, cannot power off.

is powered off.

This sequence happens in only 4 seconds.

The whole VMM thing is pissing me off. A similar error with this server on VMM 2.1 left me with a partly corrupt filesystem, I certainly do not hope that has happened again.

Must remember to remove VMM before a full VCB job scheduled for this weekend.

Ole Thomsen

[Ole Thomsen_1]

Just found this advisory

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=c01278411&dimid=933360577&dicid=alr_nov07&jumpid=em_alerts/us/nov07/all/xbu/emailsubid/mrm/mcc/loc/rbu_category_AdvisoryVirtualM/alerts

Ole Thomsen

[dave rowland_2]

What a poor response from the advisory
"reduced potential", no good to anyone
We have given up with the VMM ages ago and just use the VI Centre

By increasing the values for the polling interval delays of VM disk files in the Virtual Machine Management Agent, there is reduced potential that Virtual Machines will power off unexpectedly.

[Ole Thomsen_1]

Exactly!

Ole Thomsen

[Rob Buxton]

Here VC is the preferred way of monitoring / managing the VMWare guests.
There are still a couple of benefits for us. PMP doesn't seem to work without it on the hosts - which is a pain. And you get the guests included if VMM is enabled.

The VMM / SMP P2V seemed to produce a cleaner server when the source was a Proliant with all of the agents installed. Although some of the cleanup scripts are quite good. As we move on the P2V requirements are reducing.

So, the benefits of VMM are getting harder to justify.

[Robert Brooks_1]

So, the benefits of VMM are getting harder to justify.

--

Please wait for HP SIM V5.2 and VMM V3.1 before you give up on VMM . . .

-- Rob (part of the engineering team working on
the next version of VSE for HP SIM).

[Ervinck Kevin]

HP SIM v5.2 + VMM 3.1 are released, testing them soon...

[Ole Thomsen_1]

Upgraded to HPSIM 5.2 and VMM 3.1 yesterday.

On weekdays our VCB includes only 3 important servers, and tonight the most vital were turned off during snapshot removal :-(

Someday HP may get this right, but I am about to give up hope.

We run ESX3.5/HA/DRS on BL25p and EVA4000, latest 7.9.1 management agent.

Ole Thomsen

[Gareth H Wilson]

We we experiencing the power off problems with VCB on our HP EVA but its been fine since going to PSP8.00 and VMM3.1 on our esx 3.5 nodes

hope this helps

[AndrewBell]

We are seeing the same issue with ESX 3.5/EMC SAN

It appears the issue is no available console memory and the HA agent and other associated problems occur.

Do a TOP and ALT-M on the console to see free memory - The VMM is spawning multiple instances and consuming all conole memory causing HA to fail and the ESX host has even blacked screened requiring a cold reboot!!

To uninstall
"./installvmm791.sh --uninstall"

and stop any other HP agents. It frees up about 80MB

What is the base ESX console memory required for adding the VMM - standard is 172MB - shoudl we just up this to 220 or will the agents just spawn and consume this memory?

[Jesse Rink]

I am experiencing the VMs powering off after snapshot committal as well. Been happening intermittently for as long as I can remember. We are now on VMM 3.5 but still have the problems (ESX 3.5 U2).

Anyone had success in solving this issue? HP has been NO help so far.

[dave rowland_2]

We gave up with VMM, complete waste of time.
We now use IBM !