Running the Stack Manager on a VM


#1

I was thinking about creating a VM using Windows 2008 R2 SP1 (Since that is what I am using currently for other projects.) and installing the Stack Manager on there. Any thoughts before I go down that path?

In the long term I was thinking about creating a hosted solution running multiple stack manager instances in a Linux VM environment.


#2

It’s easy - doing it every day except my VMs run linux. Don’t over complicate it and rely on fact that this all uses ICE/STUN to tunnel back to servers, full automatic networking in domain-server settings and metaverse data setup pages are your best friends. Just try it and see - there’s no cookbook approach to it, but, if you can’t get it going self-serve then it’s likely you won’t have a good run with it over time as this code changes so rapidly that what works one way today might be very different tomorrow.


#3

Omega,

is your advice regarding the rapidly-changing HIFI code base a consideration for one to take into account when thinking about running HiFi from a cloud or other hosted solution–i.e. do the constant HiFi code updates require constant updating on the host-provider’s end as well?

I hope that is a clear question; my knowledge of how hosting works is not at full capacity, yet. I know the basics.

Thanks!


#4

And all those years I spent locking down my Network and Firewall… LOL


#5

If running on Linux - you need to be prepared to frequently re-compile you server-side HiFi provided code. Coal has made this fairly easy using Centos Linux based systems. So - you’re going to need to be comfortable being able to log into a remote server, run scripts to update code/recompile etc. There’s no “easy mode” Stack Manager or binary pre-compiled downloads for Linux. If your cloud based (or whatever) system is running Windows then it’s a bit easier if you can access the remote GUI mode.

I had pondered offering to provide domains for very close to cost of “renting” the hardware, but, for now that’s too difficult as we have no in world currency of value and I don’t want to resort to collecting L in SL for services in HF. :smile:


#6

Bingo - someone who gets it :slight_smile: Thus my somewhat paranoid response to how I run my HiFi servers - in VMs in locked down containers with severely constrained access to network/file system.


#7

This is helpful, @OmegaHeron. The servers I have access to run on Linux (actually, I think some of them run on windows), And, I think I can manage logging into the servers, running scripts and updating/ recompiling code, as long as I don’t have to write that code! (I can write some JS, but that is not going to help with the server.)

In this thread, https://alphas.highfidelity.io/t/what-should-i-run-the-stack-manager-on/3527/9?u=sterlingwright, @morrisford shared his hosted solution, which sounds very good and very easy. I would just like to understand the back-end technical (and not just user) considerations before deciding what route to take.

Someone, (is it you, Coal?) is writing up some documentation on how to host in Amazon’s cloud, which sounds like an imminently valuable contribution!

Thanks, All.

(test, test, test…)


#8

@SterlingWright - at this time @Coal has the most refined process for running these things in a semi-automated way. His posts and documents will likely be your best source of information. The only thing I think one needs to consider is what your bandwidth limits are on any given service you procure. Some of the lower priced plans have (relatively) lowish bandwidth caps and if serving models and domain data from it you could start edging up or passing limits if things get busy.


#9

I have pretty high bandwidth on each of the domains, and they are just sitting there unused.

Maybe this is a way to justify the expense of maintaining them.

I will do a full option-comparison before deciding, and appreciate everyone’s input.

Thank you.


#10

Thanks @OmegaHeron! Also @SterlingWright if you need help going through and setting up anything special, I can work with you on a consulting level to help find what works best for you.

Either way I wish you the best!


#11

Just to give an idea - I just pulled a report for my content served just to provide http content to my, formerly, public space - Heron for one week. 1.31GB. It adds up in a hurry. I still need to write a monitor to track bandwidth used by other parts of system, but, it will be a fair bit less than that used serving models/graphics.


#12

Thank you, Coal. Cool to know that you are offering consulting services on stack set-up. That should serve many people well as they come in!


#13

This begs for a Vagrant VM for content creators. That’d be awesome.


#14

@FlipperPA not saying it “can’t be done” or “shouldn’t be done” but things required change so fast around here that just doing a straight Vagrant deploy without pulling in the scripts elsewhere will make it not last long and take a new setup released fairly often. Now I could setup something like that and may in the long run.

With more people coming in I actually guess I will look for something like this.


#15

@Coal Hey, thanks for the reply and candor. Yeah, maybe once we’re past alpha. :slight_smile: It was amusing, I received the alpha invite… the same day we took our product to alpha!


#16

@FlipperPA - Nice! I love Alphas :stuck_out_tongue:

Been with this one over a year now, seems like forever!


#17

Just noticed this, and yes it is me. I just need to wrap up that document. So much other stuff has come up and been trying to keep it wrapped up before I finished that document.

I can actually tell you how to do it if you needed to discuss different options.

@SterlingWright


#18

I am curious about running the stack manager on a Droplet.