#SL ADITI Grid Problems

ADITI is the preview grid for Second Life. This is where we test new scripts, mesh items, and lots of other stuff. It is popular with creators because uploads are free. Currently people are having problems logging in. The problem is caused by the hard drives for the ADITI database filling up.

Take-Away:

There will be discussion at Thursday’s Beta Server meeting on what may be the best solution for the problem. If you use the preview grid, you will probably want to be there for that meeting.

Discussion in the SL Forum: ADITI Problems Thursday Meeting

Details:

ADITI is smaller than the main grid and runs on less hardware. The database for assets on the ADITI is much smaller, hardware-wise. Oskar Linden, no longer on staff, was handling keeping the grid running. But, the situation required manual cleaning of the drives. Now that is proving a problem.

How the problem is going to be resolved is up for discussion at Thursday’s 3:00 PM Beta Server user group meeting. See: Server Beta User Group. I suggest you attempt to log into ADITI at least by Wednesday. You may find you can only get in with an Alt-avatar.

When a disk fills up there is a finite number of solutions. While adding a larger drive seems like a simple solution, SL does not run on ‘a drive’. It uses a cluster of computers with large drives. Cost is a factor. As Simon Linden says these are not drives you can pick up at Fry’s. So, some alternative is being looked at.

The primary one is looking at how to cull out people not using the ADITI grid. Their inventory does not need to be on the preview grid. Next is the idea of flushing the inventory of anyone not using the preview grid for some time period.

Most of the solutions are going to involve reducing the number of people with inventory on ADITI. I think it is reasonable to expect that if you are an infrequent user, your inventory will get flushed from time to time.

Also, if you are someone with a HUGE inventory, it is reasonable for that flush period to be shorter. I think some incentive for you to keep inventory small on the ADITI grid is needed. While it would take some extra effort to use an alternate with a small inventory, I don’t see it as that big a deal.

I think limiting inventory to those regularly using the grid and timing out infrequent users will be the likely solution. But, if you have a stake in this decision, show up and participate.

Database

In discussing this problem information came out about how the asset system works. All the assets are not in a database. The assets are files on a disk. The asset system uses a database to index the location of the asset’s files. This makes the database way smaller.

An interesting trivia point is the asset files for both AGNI and ADITI are kept in the same filing system cluster. The databases with pointers into the system for ADITI and AGNI are separate databases.

Andrew is looking into how the ADITI grid works and what solutions may be best. I expect we may learn more about how the system works at Thursday’s meeting.

 

15 thoughts on “#SL ADITI Grid Problems

  1. Nalates
    While I understand the limitations of the aditi Grid, the idea that users with infrequent needs woulds effectively be barred from using the grid is to say the least bizarre. Time and again recently it has proven that the solutions that appear to work in the highly depopulated and artificial environment of Aditi simply do not work on the Main Grid with more users, objects and scripts. To increase the disparity between the Test and Main Gids makes absolutely no sense, if the updates to SL software are to be effective. This suggests another “knee-jerk” reaction by Linden Lab, who are now coming to regret hasty dismissals.

    • You tend to miss the points… no one is being barred or planned to be barred… nor are they doing a knee-jerk reaction.

      • Quite right, Nalates, my choice ot the term “barred” was wrong. As far as it being a knee-jerk, I guess it isn’t reflexive, but it IS a counter-productive move and will take Aditi even further away from its applicabilty in a “Pilot” role.
        Mind you after this week’s upsets one might be forgiven for wondering if the QA that is done there actually has ANY value for Agni.

        • I disagree. Figuring out how to get the people’s inventories on ADITI that need to be there and avoid loading the inventories of the people not using ADITI is very productive. I see no way that is further away from the goal of having people help with testing in ADITI.

          That anyone would wonder whether QA on ADITI benefits AGNI is surprising. I think it suggests one is oblivious to what is going on in ADITI.The number of problems found on ADITI and resolved before moving code to the release channels numbers several thousands this year.

          • I think we must agree to differ on the benefit to Agni of Linden Lab’s QA process. As regards general proceedings on Aditi, I am far from oblivious; if thousands of issues have been screened by thie process it does not speak highly of the quality of what is being tried there, or of those that are compiling said code. I know you will most likely disagree there, but nonetheless my opinion is mine and no-one else’s.

            • You are welcome to your opinion. It says a lot about you.

              I do disagree. Especially the software development process and human advancement in general is by making mistakes. That you seem to have the idea there is some process or someone or group that has an ability to eliminate mistakes and so the Lab should never allow any significant number of mistakes to make it through QA is naive. Automakers, Apple Inc., Microsoft, and anyone developing complex systems is making mistakes that reach the end user. It is life on Earth. Consider. Has there ever been a book published that has not had a misake? We have been publishing ‘books’ for nearly 7,000 years. The Lab is doing things that have never been done before.

              Last year The Gardian acknowledged that in an article titled: Acknowledging mistakes is key to advancement – and not just in science.

              Quote: If you don’t make mistakes, you’re not working on hard enough problems. And that’s a big mistake. ~F. Wikzek
              Quote: The man who makes no mistakes does not usually make anything. ~Edward Phelps
              Quote: Making mistakes simply means you are learning faster. ~Weston H. Agor
              Quote: The history of human opinion is scarcely anything, more than the history of human errors. ~Voltaire

              Read up on the discovery of metal fatigue in aircraft in the 1930’s. People died while that was being figured out. Mistakes are part of the human condition.

  2. Pingback: Aditi In Trouble » Ciaran Laval

  3. Nalates, I never said, nor do I think, that Linden Lab should, or ought to never make mistakes, despite what you think, I am not that naiive. What I DO think is that they are currently making too many mistakes and that they are slow to acknowledge errors when they ARE made. If they “held their hands up” a little more readily instead of trying to obfuscate and hide errors, I and a great many others would view them with a good deal more respect.
    And for the life of me I cannot see how making Aditi less accessible to users will in any way aid their cause.

    • I’m not at all sure why you continue to think they will make it less accessible.

      That you think you understand the error rate for a software house doing something that has never been done before and have an opinion of what the error rate SHOULD be and how they should conduct THEIR business is why I think you are naive.

      • Ah Well, Nalates
        The bottom line here is that I disagree with you and therefore I MUST, in your view, be both naiive and wrong. I think not.

        • We do disagree.

          I will point out that being naive generally means one sees things differently. That difference is usually from a lack of information by the party considered naive, at least that is what naive usually is used to denote and how I used it. I know you are missing information because you don’t work at the Lab. It isn’t that you are naive because you disagree with me. I see you forming opinions about things you can’t know enough about to justify the opinion, like the Lab’s rate of errors getting through QA. No one outside the Lab could if it is good, bad, or average.

          That you make the illogical statement that because someone disagrees with me I think they are naive, again exhibits the personality trait I associate with the uninformed (naive) or those not able to make their point on merit that adopt an ad hominem strategy. If you have read this blog you know I disagree with a number of people and do NOT think they are naive. I often disagree with Tateru Ninno on her take on SL stats and Darrius Gothly on his take on the Commerce Team. But, in both cases I respect their knowledge and opinions. We just disagree and I state why with out attacking them. So, I think your previous comment is consistent with your tendency to make harsh comments about people and groups without the knowledge and evidence to justify them.

  4. Seems to me teh solution is to eliminate the need for users to access the beta grid. Why not allow test mesh uploads to the main grid? Users of Phoenix and Firestorm have the ability to to temp texture uploads to the main grid. Add that and ability to test upload mesh to the main grid without charge and there’s no more need for users to be on the test grid.

    • The real purpose of the ADITI grid is to test new software; server versions, viewer features that need server side changes, and other things that require changes that could disrupt the main grid. To get people there to help test they provide us a free service. I think it works well for both sides.

  5. Pingback: 2012 Reviewed – October To December » Ciaran Laval

  6. Pingback: 2012 Reviewed – The Full Monty » Ciaran Laval

Leave a Reply

Your email address will not be published. Required fields are marked *