[Opensim-users] Safe use of OpenSim - was "Fw: Re: [SLED] Linden Lab Sells Second Life!"

[Frisby, Adam]

My advice would be run both.

We don't up the protocol version too often - so there's nothing stopping you from running a handful of regions at trunk, and the remainder on stable. If you have customers involved, while we appreciate the feedback - I would probably say stick with stable there.

For your own testing purposes, running trunk isn't a bad option - and it also gives you the flexibility that if something is broken majorly, you still have all your stable regions.

To be fair, the 0.6.3 release was not fantastic - the avatar bug should have probably been fixed before that was tagged. I think in future that complaint has registered pretty soundly and releases might be more thoroughly tested through release candidates.

Adam

From: opensim-users-bounces at lists.berlios.de [mailto:opensim-users-bounces at lists.berlios.de] On Behalf Of John Sheridan
Sent: Wednesday, 1 April 2009 6:32 PM
To: opensim-users at lists.berlios.de
Subject: Re: [Opensim-users] Safe use of OpenSim - was "Fw: Re: [SLED] Linden Lab Sells Second Life!"

Holy blog wars Batman!  I can see emotions are running high, so I thread lightly while writing this...

For around six months now since I and my cohorts started building out our grid we've been more or less running off of the trunk.  We have a weekly update process wherein every Sunday after the commit emails stop flying by I'll shut down, back up, then update everything to that Sunday's SVN.  After the update is complete I'll go around in world, try a few tests to make sure that everything is in one piece and that basic functionality is there.  So far there have only been a handful of times that I've had to cancel or roll back an update due to broken functionality or warnings on the mailing lists that heavy changes were taking place.

As far as data and content are concerned, we all realize that this is alpha software and that certain instabilities may exist from week to week.  So far we've not had any major disasters as far as data and content save a few mangled, non-persisting, or overly persisting (where they won't go away) prims.  However in most cases workarounds were available and all users were told how to go about utilizing them.  As an added measure, just in case we rely on the weekly backups as well periodic dumps of our sims into OAR files.

My primary reason for choosing this sort of procedure is so that while we are building, creating, and using the grid we can also contribute back to OpenSim in the form of bug reports and feedback.  (See a draft copy of our community standards for more on this:  http://www.pseudospace.net/joomla/index.php?option=com_content&view=article&id=56&Itemid=60)

Regardless, we do consider our setup to be in quasi-production regardless of whatever minor glitches exist within the trunk.  As well, we are sure to make it quite clear to the few users that we do have (albeit not many as we do not yet have any completed or usable environments other then Ellis) that both OpenSim as the underlying software and Pseudospace as the service itself are under heavy development and could be unstable at times.

Now as a software developer myself (albeit in a completely different realm then OpenSim as I do small scale database / windows apps) I can understand the risks of running untested software for production uses, however in this case I'm unsure as how to proceed.  We like contributing back to the community in whatever way we can, and also enjoy testing or riding the bleeding edge of the wave, yet after reading the prior posts I'm suddenly not sure if simply backing up, exporting, and rolling back is as safe as I thought.