just discovered bugs in update?

As I read the blog updates on the down time today, I keep seeing the same thing, "We fixed this, but we found something else broke." I was under the impression that the beta grid was where all this was tested out. If it worked there, it "should" work on the main grid. So why all of these new problems? was this update not tested out on the beta grid?

I guess I misunderstood the whole point pf having a beta grid. I thought it was where LL tested out the updates and upgrades to make sure they worked before putting them to use on the main grid.

It's just a nomenclature thing.

When they say "main grid" they mean what people would conventionally call "beta grid".

When they say "beta grid" they mean "tiny fraction of beta grid for developer unit testing where we can ignore results because they'd be a buzz-crunch".

Hope that helps.

~laughs~
so they really don't test the updates/upgrades on anything worthwhile testing!

Got it! I knew there had to be an explanation.

At least they're not giving the grid reopening time no more. As soon as the email issue is solved, they might find out the wrong high speed gerbils were used...

oh didn't you hear? the gerbils crashed from all the lag, and now, they are using rats.

SL day is shot, but maybe it will be better tomorrow! at least I got something accomplished out in the real world today.

They did try ferrets for a while, but the ferrets kept taking all the shiny things. Kind of hard to keep hard to keep the grid running when the wheel goes missing!

Better it's down NOW than up and causing us no end of problems.

The beta grid is where the Cornfield is located now...LOL.

Testing in Production is a much better and faster way to test. With 40,000+ people online, they are bound to find out about the bugs much faster than if they had to wait for a few developers and testers to find them all. Much more efficient this way - and better financially too (they have to pay those testers, where as we pay them so that we can find the bugs).

Warning: This post contains sarcasm and may cause deliterious effects on the reader. Viewer discretion is advised.

That's why they do updates every one to two weeks. They upgrade, then wait to see what happens when its live - until Thursday night/Friday morning when the bugs start crawling out of the woodworks.

A few come out at first, then by Saturday afternoon entire swarms of them infest the grid. LL then observes the bugs briefly, hits a few of them with a fly swatter late Saturday, takes notes on what happened, then says everything's fixed...

Until Sunday, when the bugs chew their way into the Asset servers. At that point, the system is borked until Monday morning when they call in the bug exterminators.

At that point the bugs go run and hide somewhere (Beta/Teen Grid?) and the system is somewhat smoother until the next Wednesday updates.

Repeat and rinse as needed.

Your mileage may vary.

Snark aside, there is a real answer, and it's true of any system that has to scale.

The beta grid is used by a much smaller number of people than the main grid. Some problems that happen infrequently may not show up at all in beta, but become an issue when the main grid is involved. Or problems that are associated with the number of users may not happen at all on the beta grid.

You see this with medicines ... a company will go through all the steps for FDA approval and still the drug causes unexpected problems when it goes on the market. Why? Because clinical trials of a few hundred people do not always uncover issues that come up when tens of thousands of people begin taking it in uncontrolled situations.

Beta testing is good for catching a lot of issues, but it won't always reveal everything.

LL indicated sometime ago, that the beta grid is being moth balled.. I guess it had something to do with Microsoft's statement about windows 95 no longer being supported.

Actually, to be fair to LL.. they now have new testing metrics, that give them better assesments to real issues under real type loads.... (now not in their defence) when they are going to start using it is anyones guess... guess they are testing it first on the beta grid <sarcasim mode on>

Phillip is now reported to be happier with the new situation.. and he can now play his 1st edition Sims programme (which the tech told him was in fact FirstLookVoiceWindlight version 3.0, running on the main grid) which has been running off the same HD shared with the original beta grid.

In case some do not know.... I met with quite a few people that have only ever used the Beta Grid and had no interest in being on the Main Grid.... oh well guess it gives someone for Phillip to play with at least.

A lot of the problems seen are things that should be getting picked up by automated testing. Other things that aren't picked up should make them go "oh! We missed that use-case; lets add it to our future tests!"

Unfortunately I get the feeling LL never planned SL, never created a canonical list of use-cases to implement, and hence never came up with any automated test routines for anything other than limited unit-testing.

You are correct in your analogy... I know personally the CEO of a company that recently completed the new software testing metrics programmes for LL to use. I'll keep my fingers crossed loosely, as I trust their abilities to supply good quality programmes, but then again, they are not the users... LL are.