BRRABill's Field Report With XenServer
-
@olivier said in BRRABill's Field Report With XenServer:
VM metadata means: VM name, description, number of virtual interfaces, disks etc. Everything except the disk content.
So when we are doing VM backup, we are actually export both disk content and VM metadata.
Otherwise, it would be just a bunch of disks with their UUID, which is a bit useless when you need to restore (unless you can memorize every VM configuration and disk placement with their respective UUID. In this case, please answer the question on the meaning on life and everything).
I just did this today on my test machine.
I did a metadata backup. Ripped out the boot disk, and reinstalled XS on a fresh boot disk. Reintroduced the SR, and restored the metadata. Worked like a charm.
It threw up one error about not being able to find one, but I think that might have been the metadata VDI itself. Stupid me didn't write down the error.
-
@olivier said in BRRABill's Field Report With XenServer:
VM metadata means: VM name, description, number of virtual interfaces, disks etc. Everything except the disk content.
So when we are doing VM backup, we are actually export both disk content and VM metadata.
Otherwise, it would be just a bunch of disks with their UUID, which is a bit useless when you need to restore (unless you can memorize every VM configuration and disk placement with their respective UUID. In this case, please answer the question on the meaning on life and everything).
But you should take a metadata backup separately, too, no?
Let's say your boot device (cough USB cough) decides to die and you don't have a backup.
It's so much faster to reinstall XS and restore the metadata than restore all the VMs through XO, right?
Though of course the XO is also important to protect from loss of the VM itself.
-
@BRRABill That's far from being the usual case. But it could happen, in this case yes it would be faster to only restore metadata.
-
@olivier said in BRRABill's Field Report With XenServer:
@BRRABill That's far from being the usual case. But it could happen, in this case yes it would be faster to only restore metadata.
Have you been following ML? Seems to happen a lot around here!
-
@BRRABill said in BRRABill's Field Report With XenServer:
@olivier said in BRRABill's Field Report With XenServer:
@BRRABill That's far from being the usual case. But it could happen, in this case yes it would be faster to only restore metadata.
Have you been following ML? Seems to happen a lot around here!
To be fair, you've done some weird things with your XS installations..
-
@BRRABill well, not in my world (more corporate world than lab env). In average, there is more SR lost or problems which need to restore entirely a VM. And if you are in a pool, losing one host completely is just a non-event.
-
@DustinB3403 said
To be fair, you've done some weird things with your XS installations..
Hasn't happened to me yet!
-
@BRRABill said in BRRABill's Field Report With XenServer:
@DustinB3403 said
To be fair, you've done some weird things with your XS installations..
Hasn't happened to me yet!
We know you can still break it!
-
@olivier said in BRRABill's Field Report With XenServer:
@BRRABill well, not in my world (more corporate world than lab env). In average, there is more SR lost or problems which need to restore entirely a VM. And if you are in a pool, losing one host completely is just a non-event.
I'm just mentioning it because in the past few months we've had it happen to at least two people here (@Dashrender and I forget who else), and since a lot of the time XS is recommended to people looking for a hypervisor that are just moving to virtualization, I just thought it might be appropriate for them.
-
@travisdh1 said
We know you can still break it!
I'm planning to migrate my XS booting off of USB and onto a RAID array so there is still a chance.
Though since it was pretty pain free to do, I might just wait until the USB stick croaks.
Nah.....
-
@BRRABill said in BRRABill's Field Report With XenServer:
@olivier said in BRRABill's Field Report With XenServer:
@BRRABill well, not in my world (more corporate world than lab env). In average, there is more SR lost or problems which need to restore entirely a VM. And if you are in a pool, losing one host completely is just a non-event.
I'm just mentioning it because in the past few months we've had it happen to at least two people here (@Dashrender and I forget who else), and since a lot of the time XS is recommended to people looking for a hypervisor that are just moving to virtualization, I just thought it might be appropriate for them.
The solution though is very simple.
Shutdown the host, and clone the USB drive.
Or if you don't want to install to USB, create two partitions on your hardware a small 32GB partition, and the rest for installation. Install and go. You then have a boot partition in RAID1, and the other in RAID10.
Problem solved. Albeit more expensively.
-
@BRRABill said in BRRABill's Field Report With XenServer:
@DustinB3403 said
To be fair, you've done some weird things with your XS installations..
Hasn't happened to me yet!
Actually.... I literally broke a usb stick the other day. Forgot it was plugged in when I was rerouting cables and popped it off the front port... The R900 wasn't even scratched.
-
You know, this might belong in its own thread, but...
I know it's a terrible idea to restore a DC. Always better to reinstall and promote.
But what about moving one from one XS to another?
-
@BRRABill You mean just migrating a running vm that is a AD DC? Done it many times.
-
@BRRABill said in BRRABill's Field Report With XenServer:
You know, this might belong in its own thread, but...
I know it's a terrible idea to restore a DC. Always better to reinstall and promote.
But what about moving one from one XS to another?
I'm so happy in the Linux world. Not trolling about MS products, just facts about fragile some windows VMs can be when there is a problem.
My clients experiencing the VDI corrupt stuff:
- 3 Windows VMs, all corrupted
- 6 Linux VMs, some with database or web server: no problem
Can be a complete random (very small number to draw any conclusion), but still. That's very common.
-
@momurda said in BRRABill's Field Report With XenServer:
@BRRABill You mean just migrating a running vm that is a AD DC? Done it many times.
Or shutting it down and doing a full move.
-
@olivier said in BRRABill's Field Report With XenServer:
@BRRABill said in BRRABill's Field Report With XenServer:
You know, this might belong in its own thread, but...
I know it's a terrible idea to restore a DC. Always better to reinstall and promote.
But what about moving one from one XS to another?
I'm so happy in the Linux world. Not trolling about MS products, just facts about fragile some windows VMs can be when there is a problem.
My clients experiencing the VDI corrupt stuff:
- 3 Windows VMs, all corrupted
- 6 Linux VMs, some with database or web server: no problem
Can be a complete random (very small number to draw any conclusion), but still. That's very common.
I still don't understand why AD isn't smart enough to just fix itself.
-
@BRRABill It's a monolithic/packaged/all in one solution, with a lot of layers. So it makes sense in a way.
-
Try moving your vdi while vm is running, to another SR, very exciting as you dont know if you will get the dreaded, 'The VDI mirroring cannot be performed" and your vdi just disappears. Luckily it is just detached and back in its original SR. Seems to only happen with windows vms for me
-
@momurda said in BRRABill's Field Report With XenServer:
Try moving your vdi while vm is running, to another SR, very exciting as you dont know if you will get the dreaded, 'The VDI mirroring cannot be performed" and your vdi just disappears. Luckily it is just detached and back in its original SR. Seems to only happen with windows vms for me
Ouch. Tools installed I suppose. This error mean tapdisk failed to write new blocks to the destination
(XenServer issue)
