How much can we relay on HP VMs on Integrity servers

[senthil_kumar_1]

Hi All,

I am new to HP-UX.

I have configured HPVM on Integrity server 3660.

Have configured 5 VMs on that.

HP-UX 11iv3 has been installed on host and VMs.

One suddenly I have restarted host server mistakenly while all the VMs are running.

After the reboot of host server, one of the VMs is not booted, it is going upto EFI only.
and i am not able to boot in to HP-UX OS installed on that VM. After doing some R&D, I have found that all the data (including boot image and OS) have been lost due to inproper restart of VM.

It is not put me in big trouble since that VM is still in testing phase. so we have installed the OS newly. and now we are using that VM.

Virtual LV disk (one LV of host) has been used as disk for VM.


I have some following important questions:

1)How to save the VM in future from such incidents?

2)Why it was happened?

3)Is HP VM not recommended for production server?

4)Is it possible that we can boot by old data if this situation comes again after converting it as production server?


5)Is it possible that backup and restore the whole VM to older state like snap shot facility available in VM ware?


6)Could you please tell what are other precaution steps we need to take while using HP VMs?

[smatador]

Hi,
I have quite same config rx3600 on R&D with 4-5 VM.
1) I save them by ignite tools like make_net_recovery, I ever save one on tape and the recover was successfull.
2) I don't know what happen for you. For me, I test HPVM since months, I do very nasty things and I'm very surprised because VM start every time. For example I delete the efi partition, recover from the alterned, always good. Perhaps, you made some command that corrupt the OS, pvcreate in the host ?
3) I have some customer that use HPVM in production server.
4) Always made a backup. You could also in case the guest are in device disk, stop the guest and made a dd to another disk.
5) I'm not aware of this type of features.
6) In the doc, all are writed. So I will just add
- save the command you do to create the hpvm
- always have an output of the hpvmstatus.
HTH.

[Johnson Punniyalingam · Accepted solution]

>>>1)How to save the VM in future from such incidents?

Please Setup "Ignite Server" HPVMHOST and all other VM Guest has your Ignite Clients

>>>>2)Why it was happened?

Why did you shutdown accidentally ? Server . We should be always careful while doing our job
offcourse, " imagine if human run very fast on the road .!! Suddenly you drop Big Stone on his Head. off course everything will panic that's what happened. its always better to "have Ignite backup" of the clients

>>>>3)Is HP VM not recommended for production server?

simple answer, Yes for me "because so many "Organisation going into virtualization technoloagy has long has you have configured your (production servers)in consideration of avoiding all SOP -> Single point of failure, I think your safe you can run production server in "VM"

>>>>4)Is it possible that we can boot by old data if this situation comes again after converting it as production server?

Yes, You can, as long as you have "VM" Ignite image backup


>>>>5)Is it possible that backup and restore the whole VM to older state like snap shot facility available in VM ware?

Yes, You can "Ignite" VM client OS from Ignite Server (VM Host)


>>>6)Could you please tell what are other precaution steps we need to take while using HP VMs?

properly planning , to avoid all Single point of failure

* Disaster Recovery planning -> like you can also cluster "your VM guests" to fail over to another node,

*Setting up proper "Backup Solution" Incase of recovery the OS and Your application data

* Aggregation your "lan card" by dedicated lan card for you "VM Guests"

Hope This Helps

Regards,
Johnson

[senthil_kumar_1]

Hi smatador,

I am new to HP-UX.

1)Is rx3660 having tape drive with it by default for taking backups?

I am not able to see any tape drive using the command "#ioscan -funC tape".


2)Do we need backup what like host or all VMs?

[senthil_kumar_1]

Hi Johnson,

Could you please give any docs for installing and configuring "Ignite server on host" and "Ignite client on all VMs"?

[Johnson Punniyalingam]

Sure >> No problem>>

[Johnson Punniyalingam]

Check Below Thread

http://forums13.itrc.hp.com/service/forums/questionanswer.do?admit=109447627+1264439344723+28353475&threadId=1288879

[Steven E. Protter]

Shalom,

Clarification:

make_net_recovery -- writes to tape.
make_net_recovery -- writes to an NFS share.


Use the appropriate tool to back up you system before you mess with your system.

What document did you use to set up your VM environment?

How many CPU's are there?

How much memory?

You may have a problem starting VM because there isn't enough memory.

What OS, 11.23 or 11.31 is in use here.

SEP

[Olivier Masse]

1 --
make_net_recovery

2 --
I have no idea

3 --
I've been administering ESX and HPVM servers for a while and the first thing I've learned (the hard way) is that when you put too much VMs on a single host and that host experiences a hardware failure you're in trouble.

So... if you run production on VMs, I suggest you have multiple clustered hosts with online VM migration.

4 --
Possibly if booting off a SAN and using raw disks backstores, but not easily as a VM won't like being rebooted on a physical server and you'll need to fix vg00. make_net_recovery is the easiest way.

5 --
There's no easy mechanism I know of to do this as in ESX. You could possibly use VxFS backstores and make VxFS snapshots, but that's pulling strings a little too much and performance will suffer. The easiest way is to use a SAN and use raw disk back ends for boot and data, and snapshot directly on the SAN.

6 --
a) Try to keep up to date with the VM CORE patches. This will save you some trouble.
b) Double up everything on the host (HBAs, network teaming, power supplies ...)

Olivier.

[Emil Velez]

IN addition virtual machines can be made highly available and somewhat protected from a server failure by making the virtual machines serviceguard packages.

It seems that you need to do some planning for contingencies before using this technology.

[smatador]

Hi,

1)Is rx3660 having tape drive with it by default for taking backups?
I am not able to see any tape drive using the command "#ioscan -funC tape".
> I use an external tape drive, for example I have an HP C5683A attach on a pci slot with a SCSI Ultra320 A6961-60011 card.

2)Do we need backup what like host or all VMs?
I do VmGuests and Host backups. VMGuests because there are the productions servers, so if you loose one you could restore quickly.
If all your VM are the same [ vg00 ] and you have some trouble with disk space in you ignite server, you could clone one VM to another and maintain only one with multiple ignite config file corresponding to each VM. Like that you could reinstall VM and vgimport vg data.
HTH

[WayneHP]

Sentil,

You have asked over 2000 questions since March 2009? Is this right?

http://forums11.itrc.hp.com/service/forums/pageList.do?userId=WW220552&listType=question&forumId=1