Michael Haren's questions

Suppose I want to scale up/down my Azure Cloud Service Programmatically. The scaling up case is easy. But when scaling down, can I control which instance gets torn down? Is this predictable?

In my particular use case I have a queue of very long-running operations that I fan out to multiple instances based on some criteria. Each operation might take as long as 20 hours, or as little as 2 minutes.

If I use Wasabi or auto scaling, it looks like random instances are selected for tear down (or usually the most recently added instance...but I can't find a guarantee of that). This won't work particularly well for me. For instance, I could spin up 20 workers to process a queue of 20 work units only to have worker #20 stuck with the 20 hour job.

Is there a way to choose which instance will be torn down (either from within the instance, or from a dedicated management node)?

Suppose I spin up a machine in EC2. Every time I stop/start the machine the following things change (reboots are safe--they don't cause these to change):

• public IP address
• public DNS record
• private IP address
• private DNS record
• removes associated elastic IP

How can I reliably connect to the machine if the addresses/DNS records change every time it is started?

Should I stick with EIPs but add a startup script to reassociate it with my instance when it boots?

There's got to be a better way!

What's a good way to benchmark a network connection? I'm looking for something like the Windows Experience Index, but for my internet connection.

For example, one place I visit has what seems to be a terrible connection. Pages load slow, etc. but file downloads are insanely fast once they get started (bandwidth good, latency bad?).

Alternatively, while tethering through my phone, browsing the web is nice and brisk, but downloads are excruciatingly slow (bandwidth bad, latency good?).

What's a good way to evaluate these variables (bandwidth and latency?) and are there others (load?) I should consider?

My LDAP users aren't showing up in Subversion Edge's web console. How can I add them to existing roles if they don't show up?

I am setting up CollabNet SVN Edge. So far it's great but I'm wondering how I should access the admin functions from another machine. While connected to the server, I see I can access the remote console at:

https://localhost:4434

Is the recommended approach to simply open up port 4434 on the machine's firewall and use that same address (with the host name, obviously)?

Or should I be getting to the admin area another way?

I have two NICs:

1. Network 1
2. Network 2

Network 1 is awesomely fast but blocks access to Site A. How can I route traffic to just Site A through network 2?

When I have them both connected, Windows somehow knows to use network 1 for everything (I assume this is via the metric setting?).

Site A can be an IP address (vs. a domain name) if it makes this easier.

I infrequently work on about 20 Windows servers via remote desktop. There are times when it'd be really convenient to know if and where I have disconnected sessions open.

Is there a powershell script or query or something I can use to make this process simpler than logging into each machine and checking Task Manager's users tab?

I have an intranet site using Windows Authentication. When users are prompted for credentials, they enter:

User: domain\username
Pass: SomePassword

It'd be nice if they could leave off the domain\ part which would be filled in with a default domain. I'm willing to give up the ability to login with a server/machine account (non-domain) because we don't use that anyway.

This seems to be possible with BasicAuth but I can't seem to find a reference for how to do it with WindowsAuth.

TL;DR:

flow chart http://img2.pict.com/6e/d6/a7/2101822/0/mockup.png

I'm curious about how email flows into the Droid handset from my corporate MS Exchange server. If I run through the regular Exchange config wizard and configure it to talk directly to my company's server, does it flow directly from my server to my device or does it still (somehow) route through a third party first (e.g. Google or Verizon)?

Also, how can I tell if it is securely transmitted (e.g. SSL)? Is this all the same when sending mail from Droid, too?

+1 if you have references!

I am attempting to run Redmine (RoR) through an Apache reverse proxy. Mongrel serves plain-text pages which Apache protects with SSL and serves to my users. The problem is that various links and redirects within Redmine send users to http:// instead of https:// .

The settings on the admin page are configured for https but this doesn't seem to help. I also added this line to environment.rb to solve a relative root problem (that fixed it, but not the https problem):

ActionController::AbstractRequest.relative_url_root = "/r/" 

How can I force Redmine to always use https?

Note that a redirect won't work because my server only has :443 open and thus :80 requests wouldn't be seen to be rewritten. A rewrite of the site's data could work but seems less than ideal.

I'm replacing a sever and wanted to know if there's a way the new server can have two names. For example:

What we have now:

Old server: srv
New server: srv-new

What I'd like to have after we cut over to the new server:

Old server: srv-old
New server: srv-new, srv

Ultimately I want to phase in the new name ('srv-new') and abandon the old name ('srv') but I'd like both names to work for a while.

Do I just do this in active directory somewhere?

We're using Active Directory. srv-old is Windows 2000 and srv-new is Windows Server 2008. Please pardon my possible misuse of terminology--I'm on the fringe of my comfort zone!

If I have network traffic between two VMs on the same Hyper-V host does the traffic actually ever leave the host server? How can I tell?

Two basic questions here:

If I want to serve SSL content from 3 different servers (A, B, C) through a reverse proxy (P), where does the SSL magic happen?

I'm guessing that each server will be responsible for handling the SSL stuff and X just routes the bits, but I'm not sure. Presumably in this case, I would use the same cert signed for X on each machine A, B, C. Or does X do all the SSL stuff by itself?

Second, what's the best webserver to use for the reverse proxy on Windows Server 2008? IIS, Apache, etc. Extra credit if you have a good example of setting one up with your recommendation.

Thanks for your help!

We've got a big Hyper-V host machine running Windows Server 2008 Datacenter. This hosts all of our VMs.

I'm about to create a bunch of Server 2008 VMs for various projects and wanted to know what edition I should install. Our license allows us to use whatever we want so cost is not an issue.

Are there disadvantages to using Enterprise or Data Center?

We don't really need anything beyond Standard Edition today but if using enterprise or data center doesn't hurt anything (and in our case doesn't cost any more), I might as well use those, right?

We're running Hyper-V on Windows Server 2008. I have a production environment VM for which I'm planning some major application upgrades. These upgrades are completely within the VM (not at the host level) and are not related to the OS.

I'd like to clone the production VM so I can test the upgrade procedure without affecting active users.

So my questions are:

1. what's the easiest way to clone the production machine?
2. When I do the actual upgrade, is a VM snapshot a good way to make a just-in-case backup? What does a snapshot not protect me from?

While updating my gentoo linux box, I managed to mess up PAM. Now I can't login!

Here's the message I get

login: <I type my name, enter>
/bin/login: error while loading shared libraries: 
              libpam_misc.so.0: cannot open shared object file: 
                No such file or directory

login: <repeat>

I think I need to boot in single user mode but how do I do that?

Don't worry about PAM--I'll figure that out once I can get some control of the system. If I need to use a live cd, I'll have to wait until tomorrow as I don't have any handy.

Update: I tried adding to grub's kernel line

init=/bin/sh real_init=/bin/sh

and that helped. Now I get dumped to a shell pretty much right away. The problem now is that my keyboard doesn't work! It's USB...

Update: Adding these to the kernel line also don't work:

init=/bin/bash real_init=/bin/bash (no keyboard)
init=/bin/bb   real_init=/bin/bb   (kernel panic)
single (no change--original problem persists)

I'd like to create a script to do various things when the network connection changes. For example, if the machine connects to my office network I might desire it to:

• mute itself
• change its DNS
• change its proxy settings
• change default printer

Then, when the machine is off the office network it might revert some of those.

Presumably the "active network" would be based on IP address/gateway range or WIFI network name, etc.

Do you have any recommended tools for accomplishing this? I can see something like this being applicable to many IT pros so I'm sure there are some good tips out there.

Note that this is desired to function when the network changes--not just when the user logs on/off.

I formatted an external USB harddrive while it was connected to a 32-bit Windows 2003 Server Std. edition server.

After loading it up with files, I moved it to my Windows XP SP3 where it didn't show up automatically in My Computer. I opened up Computer Management>Disk Management and see it listed as a "Health (GPT Protective Parititon)".

What's up with that? Can I mount it?

My Windows Server 2003 Std server refuses to server ASP.NET content. It serves regular html just fine but anything .net, even a one line html file with an ASPX extention fails silently.

Things I've tried:

• Nothing in the event log or IIS WWW logs when it fails.
• Fiddler shows no response
• I reinstalled .NET with
  • C:\WINDOWS\Microsoft.NET\Framework\v2.0.50727>aspnet_regiis.exe -U
  • C:\WINDOWS\Microsoft.NET\Framework\v2.0.50727>aspnet_regiis.exe -I
• I give obscenely high permissions on everything I can think of (full control, read, write, etc.) to all possibly relevant users (IUSER*, ASP.NET, etc.).
• I confirmed that ASP.Net v1 and v2 Web Service Extensions are "allowed" in IIS
• Confirmed that the Server Manager had IIS and ASP.Net roles enabled

Again: this is the scenario:

• http://localhost/Test/Default.htm <-- Works great!
• http://localhost/Test/Default.aspx <-- Bombs silently with no message at all

Any guidance will be much appreciated!

Solution: I reinstalled per the instructions below and it works now. Thanks all!