@scottalanmiller said in Always Virtualize Domain Controllers:

@black3dynamite said in Always Virtualize Domain Controllers:

It would help if Microsoft would also recommend to always virtualize domain controllers.

They do. They've been really clear on that from everything that I have seen.

Last recommendation seen on microsoft official docs - maybe the italian ones: virtualize the AD first instance, keep a phisical one as second instance...

don't understand the logic of this. but hey: their offical business support said I was not allowed to virtualise more than 2 VMs on hyper-v, even if there were linux 😕

@obsolesce It's called paravirtualization. And Xen has done it since the 1990s 🙂 Only Xen could do it for the entire VM, not just the drivers.

I should add that this is not exactly true. "Windows Dedupe" is not available in Hyper-V, which might sound obvious, as it is a Windows feature. But Hyper-V does have ReFS which has a new feature in 2016 that is not called dedupe but acts like a really light dedupe option. So while "Windows Dedupe" is not available, there is a dedupe technology available to use. But it's not a great one, it's super basic. But it also has like zero overhead, so can be pretty nice to use even if it doesn't do a lot.

@FATeknollogee said in Korora 25 Cinnamon: watercooled workstation build:

@StrongBad said in Korora 25 Cinnamon: watercooled workstation build:

@FATeknollogee slick

http://www-rhstorage.rhcloud.com/blog/vtrefny/blivet-gui-20

I just face palmed reading that first paragraph:
New features
LVM RAID support
One of the biggest features of blivet-gui 2.0 is support of LVM RAID. When adding a new LV, you can now choose RAID level for it. LVM supports the same RAID types mdraid does but you can choose different level for every logical volume so it's more flexible than using LVM on top of an mdraid.

LVM still uses md raid, just like you would normally.

That said, people like their gui for all the things, why not have yet another tool to manage everything.

@Tim_G said in Caching Needs and SSDs:

you have something on I/O path going wrong ;(

What do you mean? Everything looks good and appears to be going smooth according to all monitoring. No errors or performance issues.

It's been doing a very good job the past couple years. I keep it fairly updated. So if something recent broke something on it, let me know and I'll hold off on updating.

@Tim_G said in Sage 50 Quantum in Hyper-V VM:

So when they are ready to make their software production-ready, it will be the other way around.

Exactly. Like if they ONLY supported VMware or ONLY supported Hyper-V or ONLY supported Windows we'd understand. Those are all potential production ways to run software. But a physical deployment? This means that they've never tested and don't support ANY production environment.

There is a big gap between having "one tested environment for production" and "not tested or supported for production."

Exact same reason we dropped FogBugz. They only supported non-production platforms.

@RamblingBiped said in KVM Installation and VM Creation on Fedora 25:

I did an article about a year ago about setting up KVM/QEMU on Ubuntu 16.04, it might be of some use: https://www.ramblingbiped.com/build-a-kvm-qemu-hypervisor-on-ubuntu-16-04-server/

I also wrote a short blurb on automating VM creation using a pre-built VM template, virt-clone, and virt-sysprep: https://www.ramblingbiped.com/build-a-kvm-qemu-hypervisor-on-ubuntu-16-04-server/

That is very helpful. Thank you.

@hobbit666 said

I still love XenServer and XenOrchestra as my one of choice for now. But then i'm only running small Linux/Windows servers on it anyway like Zabbix, Radius, Unifi. Nothing full production like the DC or Dynamics/SQL

Same here, but cautiously watching...

@scottalanmiller said in HTML5 management interface for KVM:

Calligra was always okay. Not great, but certainly not bad.

0_1495832467386_Screenshot from 2017-05-26 16-00-50.png

That is atrocious

@Tim_G said in How to Download Hyper-V 2016:

Yay! Finally something real to link to.

That's why I made it, I needed it this morning 🙂

5.8.1 was release earlier today. My test build is failing, but haven't had time to dig in and find out why.

I totally understand workload being a legit reason. It intrigued me to, but to only have a week to complete some of those designs and work a normal job is rough.

@Dashrender said in Cameyo - Application Virtualization:

@RojoLoco said in Cameyo - Application Virtualization:

Didn't @art_of_shred recently get a Cameyo?

I don't know about recently 😛

Maybe at MangoCon it can make a Cameyo appearance!

Thank you for your feedback on 5nine Manager, guys.
Our goal is to provide an intuitive yet powerful product for complete centralized Hyper-V management experience.
It is now de-facto standard for managing environments with up to 20 Hyper-V host.

Now we are going even further, and prepared a great product, that can help you scale, segregate user management roles with RBAC, automation support and even more. Enjoy!

Beta program for 5nine Manager Datacenter is now open to public:
http://try.5nine.com/beta-5nine-managerdatacenteredition/

@coliver said in When You Think That You Need a Physical Server...:

@wirestyle22 said in When You Think That You Need a Physical Server...:

@JaredBusch said in When You Think That You Need a Physical Server...:

@EddieJennings said in When You Think That You Need a Physical Server...:

@coliver said in When You Think That You Need a Physical Server...:

@EddieJennings said in When You Think That You Need a Physical Server...:

I know there is a best practice that discourages an environment with only one domain controller.

Why? Do you really need two domain controllers? How many authentications are you doing? How much downtime can you afford? Would it be better to have a single domain controller on a VM that you can backup and restore in a few minutes versus having two running at all times?

Why = because a document from Microsoft said so and at the time when I made our domain I didn't know any better :).

What you're asking me is what I'm asking myself, which moves me to the conclusion that when it's time to make the VM for the accounting software, the old box should just go away. Especially since my tiny number of users would be able to log into their workstations with cached credentials until I can get the domain controller VM functioning again.

Who cares what some paper from the company selling you the licensing says.

What does your company need?

I have never used two domain controllers in the SMB space. Even before virtualization at my clients.

It is simply not something needed.

You don't think the downtime justified the cost for a SMB I'm assuming and load balancing isn't a concern

Rarely is downtime worth the cost of mitigating it in an SMB environment. They often don't actually understand what the true cost of downtime is and exaggerate it more often then not. If you're getting enough requests that you're hitting a performance threshold on the domain controller then you may be out of the SMB space.

And authentication often has a near zero impact for short durations. A DC down could easily go 30 minutes and literally have no one notice.

@Kelly yeah the HC3 Move (powered by Double-Take) is certainly easier and more direct...

if the xenserver export virtual disk file is a VHD file, you can probably skip converting it to qcow2 but simply renaming the file to match the guid of the exported empty qcow2 file (so it will still be named <guid>.qcow2 even though it's actually your VHD file.) Our import is actually using the qemu-img convert tool which will automagically detect that it's really a VHD format and do the conversion at the same time it imports it to save that extra step. (and also you can use that same "import shell" over and over... just keep replacing the guid.qcow2 file with the one you want to import)

I just write an article about that here: http://www.francescoprovino.com/2017/04/16/full-disk-encryption-unlocking-from-virtual-serial-tty-paranoia-xenserver-and-libvirt/ .

and after a lot of episodes, I've fixed the thing.

please let me state this again: I HAVE.

I.

Not the HPE support, not the reseller tech support.

I HAVE.

I = An almost idiot ex-embedded sw developer illogically cast into the sysadmin role of a non-sense company!

TL; DR:
there was a mismatch between OS version, bios version, controllers firmware version, controller drivers version. Selecting the right combination has (apprarently) fixed the issue!

how storage controllers affect reboot signals is still a mistery... but actually all was related to the smart paging I mentioned before. Moving it from the default location caused all the issues.
I have discovered the "bug" recreating a new vm with copy-pasted hdd but with hypervisor defaults... 2 twins VM one running one not... only difference: location of the smart paging file. XD