Current Time: 15:26:04 PDT

What's Up?

If you are experiencing a problem that has not been reported here, check our web panel for more information.

(Please remember, posting in the comments here IS NOT an official way to contact DreamHost.)

Search

Pages

Categories

Other Stuff

6:11 pm

Server Architecture Migrations

Posted (February 1st, 2010 at 6:11 pm PST) by Patrick

Over the past two years, we have been using a network image to boot all newly deployed servers. This allowed us central management of our operating system image, as well as an easy upgrade path. Unfortunately, within the past few months, the software that we have been using to do this has become terribly unstable. As a result, over the next several weeks we will be converting machines back to a more traditional setup (where the operating system lives locally).

What this means for you:

Every server that is currently booted over the network will be rebooted at some point to come up on a local system. That’s it! It’s a fairly painless process that, to the end user, is nothing more than a typical reboot. Sites will experience roughly 5 minutes of downtime while the servers come back up. All conversions will occur after 6pm PST and be posted on dreamhoststatus.com.

Thank you for your patience and if you have any questions or concerns, please contact support.

The Happy DreamHost Netboot Slaughter Team

This entry was posted on Monday, February 1st, 2010 at 6:11 pm and is filed under General Outages, Multiple Server Issue, System Changes. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

11 Responses to “Server Architecture Migrations”

Wooooooooooo. Thanks!

Thanks for the forewarning.

I was wondering what was going on after I made a change.
I had no idea this was going on but luckily I thought to check this.

@Scuba:

Any change you made to your account should not have been affected by any of these conversions. If something you did was changed or reverted, please contact support with details.

Be forewarned. We were moved and were down. After several hours of trying to figure it out, they moved us back.

I’m on Eastman. After the reboot, all my FastCGI sites stopped working. CGI::Fast module has disappeared, along with ghod knows what else.

I installed my own local copy of CGI::Fast from CPAN and updated my scripts with ‘use lib’ to point to the local directory. The scripts at least compile now with perl -c, but I’m still down. Now, I’m getting some strange permission error:

[Fri Feb 05 18:13:25 2010] [error] [client xx.xx.xx.xx] suexec failure: could not open log file
[Fri Feb 05 18:13:25 2010] [error] [client xx.xx.xx.xx] fopen: Permission denied

WTF?

Everything was working perfectly up until the reboot 22 hours ago. This is extremely annoying.

Is it too much to ask to do a little testing before you nuke people’s servers? Or at least spider the sites on a box before and after to make sure everything came back up? It would be a trivial script to write.

Well, I managed to fix everything, no thanks to you guys.

- Perl moved, libraries disappeared, I had to patch scripts all over the place under multiple accounts to get things working again.

Thanks a lot. That’s a couple hours of my time I’ll never get back… or get paid for.

The advantage of network-booted operating systems was that you could centralize and standardize configuration.

Will you still be able to do this with locally-installed OS’s? Or will each server become an island unto itself, each Dreamhost user having a slightly different experience?

@Joe Pepersack
Where did it move from and to, and which libraries disappeared?

 
© 1996-2010, DreamHost.com
Entries (RSS) and Comments (RSS).