OSGrid Going with FSAssets

Hypergrid Business has an article up about how OSGrid has left RAID MySQL behind to go with FSAssets, a high availability system of data storage.

my inner light
my inner light by j o r.d a n, on Flickr

It is all kind of geeky. If you are into that stuff, see HB’s article: Avination helps save OSgrid, donates cluster storage solution.

For the none geek, RAID and FSAssets are different ways of preparing for hardware failure and safely storing data. RAID stands for Redundant Arrays of Inexpensive/Independent Drives. The idea is if you have two hard drives with the second being a copy of the first, you’ll still have a good copy if one drive dies. It gets way more complicated as speed is another facet of RAID. But, two drives being safer than one is the base idea. 

Read more

OSGrid Up 2015-09

Over on Hypergrid Business Maria Korolov is reporting: OSgrid is back.

New Firestorm Region in OSGrid
New Firestorm Region in OSGrid

February 24, 2015 @ 6:45 PM the OSGrid folks tweeted the re-opening of the grid. Plus the blog has an announcement: OSGrid Online.

Yay! While this doesn’t mean OSGrid is completely up or stable, it is open to logins. They warn that there may be outages as servers are tweaked. But, they feel the grid should mostly be up and open and no more major problems.

For those that run their own regions, new server software is available for download. Also, there are some instructions in the OSGrid blog for getting existing regions back up. See the announcement link above.

As you might imagine, there is likely to be a rush of people logging in. This may place heavy load on the system and slow things down.

OSGrid Assets Recovered

Yesterday a new post appeared on OSGrid News: osgrid update for 2015-02-16.

New Firestorm Region in OSGrid
Firestorm Region in OSGrid

The assets have been recovered and loaded into the asset servers. Now the network configuration is being setup to put the servers online. There is no ETA for completion of this work.

There are discussions ongoing about how the OSGrid will be restructured. They say no drastic changes. But, there is some plan to modernize some aspects of the grid, whatever that means.

The next update for users is planned to be 2/22 or 23.

OSGrid Needs Help

You may have heard that OSGrid had a hard disk crash. They use a RAID (Redundant Array of Inexpensive Disks) system to improve reliability and speed. But, the system crashed and could not be recovered as they thought would be possible.

They have been exploring recovery via professional data recovery services. It appears they will be able to recover all data. It is unclear to me exactly what the cost is, but US$2,600 is close. That uses up a little more than half the funds available to OSGrid.

If you use OSGrid, it is definitely time to kick in with a little extra support. Make donations here: OSGrid Donations. A thousand people need to contribute $26 to cover the drive failure.

Read more

OSGrid Down

Hypergrid Business has an article up on problems over at OSGrid, my favorite OpenSim grid. It seems that Tuesday or maybe Monday some time the asset servers went down. The word coming from OSGrid peeps is that the RAID drives in the asset system failed.

The content of the asset servers will have to be restored from backups. There is typically a huge amount of data in an asset system for a virtual world. So, this will take some time. They are giving no estimates of when they think OSGrid will be back online.

Not knowing how big the data collection is for OSGrid or what the backup media is or how it is connected we can’t guess at the time needed to move the data. But, it could be days, hopefully way less.

OpenSim with 1200 Avatar Regions

That is the goal. Justine has an article up providing details on the latest testing. See: OpenSimulator Load Test 5 Report. The current tests are up to 400 avatars, which I think is pretty awesome.

It seems the new Bullet Physics is much more efficient than ODE. But, it is currently running as a single thread. This means that when they see frames rates dropping it isn’t from CPU overload. The bottleneck is in the Bullet Engine. Changing to multi-threaded would use more of the CPU and help keep frame rates up. But, changing to multi-threaded is a significant task and will be handled somewhere in the future.

They do load testing on Tuesdays. They need more human testers. Testing happens on the Conference Grid.