Questions (and answers) from our Omaha TechNet Event

  • Article

We had a great time in Omaha!  Thanks to all of you that came to our TechNet Event last Thursday. 

As usual, I had a few questions that I wanted to find the answer to so that I could put them here in my blog.  Here they are:

----

“In Windows Azure you say that the fabric controller watches for what servers are running, and can recover a role instance if it sees that the server is down. But what assurance do we have that the fabric controller won’t go down?”

I am told that you can check the XML file and look for the setting:  <disk_merge_pending type="bool">False</disk_merge_pending>

Beyond that I wasn’t given any details.  I hope that helps.

“I had a virtual machine with a hard disk on a running server that ran out of space. I needed to expand it, but couldn’t do it while the machine was running, and then couldn’t because there were snapshots. I needed to delete the snapshot and let it merge first. The merge took quite some time - which my customers didn’t appreciate at all, since I am hosting web services for them. What should/could I have done (if anything) to avoid this?”

Well.. obviously the first thing you do is always to consider treating your virtual hard disks as having a definite limit that you need to watch.  Yes, it’s nice that you have the option to expand the disk, but as you saw, there are limitations as to when you are allowed to expand a disk.  So the best approach is to monitor the disk usage just as if it were a physical piece of hardware. 

That said, the people at Microsoft who are responsible for Hyper-V definitely know about these issues that some people are having with snapshot-delete / offline-merges.  They won’t tell me whether they’re addressing this in future versions, however.  I do hope they are.

 

“I want to build servers and deploy Windows Storage Server, but I’m unable to attain that software, because I’m not a big enough reseller. Is this something that may change someday?”

Unfortunately I was not able to get a reply from anyone responsible for making this decision.  Very frustrating.