I’m no networking expert, nor a Second Like guru but I do understand some basic concepts as to why “bakefail” occurs.
Second Life servers suck. 🙂
When Second Life was being developed, they always had the idea of having different clothing layers consisting of diffrent textures. You have your avatars skin, the “original” tatto layer that was turned into being a skin layer that we know now (and then we got a diffrent type of tatoo layer), plus underware, shirts, jackets, underpants, trousers, shoes socks and so on. In recent years they’s allowed us to wear up to five layers of the same clothing type. So now you’re looking at about a two dozen textures you have to download, and render for the appearance for your avatar’s clothing.
If you think about it, if your PC and everyone else who is near you had to handle all those textures, plus everyone else dealing with everyone else’s textures (being roughly 1mb size each) performance would drop to a standstill. So the idea of a “bake” is to simply meld all of those textures into a single texture (actually three textures head, torso, and lower body). It’s just like in Photoshop when you “flatten” an image by putting all your layers into a single layer image. So now you only have a few textures to download and render, instead of a few DOZEN consisting of several megs of data, to render an avatar’s clothing.
Now how Second Life does is that this “flatting image” or “bake” is done on the user end. No you’re not “baking” everyone’s textures you see near you; just yours. Just as everyone else standing around near you has “baked” their own clothing textures. So basically what your viewer does is download all the relivant textures you’re wearing to you, and you meld them all together into three images that your viewer uploads to the simulator, and then the sim sends it to all the other avatars who are in your viewing distance.
Simple right? So why does it keep fucking up?
Problem lies often with LL’s poor infrastructure and how their asset database is constantly getting hammered all the time causing data requests to just go unanswered. It’s not really a matter of how fast your internet speed is, or how reliable your ISP is. It’s just how congested LL’s internal private network (vpn) is. Often it’s not really a bake fail but a failure for your bodyshape to download to you, and if you don’t have a shape your clothing will not render (not even sure it bothers to bake textures if your shape isn’t loaded). Also re-baking doesn’t trigger your avatar’s shape to reload either. Often just changing your shape even if it’s a copy of the same shape (not a link) will fix your shape, and then you can put on some clothes.
The behavior of the SL simulators and the asset server it self SEEM to be that when you have a bakefail, the servers assume you DID get your textures and you’re okay when you’re obviously not. This is why re-baking doesn’t always work since I do not think the act of re-baking (ctrl-alt-R) re-requests your textures from the servers anyway… it just assumes you already have them. Often you MAY have them and you have baked your textures but for whatever reason fails to upload to the simulator. That COULD be a networking issue on the users end, but if it was then it would occur consistently.
The Phoenix/Firestorm support team basically recommends to do the following to fix a bakefail, cloud, or “ruthed” avatar. And also when you get errors referring to how you can’t wear anything because your shape hasn’t loaded.
Make a copy of the following items (NOT LINKS) into their own folder.
- Body shape
- Avatar Hair (usually the balled hair set not your prim hair)
Now next time you’re all messed up, simply right click this folder you made with these items and “REPLACE OUTFIT” and this should fix you being stuck as a cloud, or a default Ruth shape. Bascially you can use and shape, skin, hair, and eyes you have laying around. The ones you find under the Library folder in inventroy will work fine. You won’t look like how you want to be but once you use it you’ll be allowed to change into your normal lookset. As long as ANY shape loads properly you can then change it.
Linden Labs recently announced a way to fix bakefail issues called “Project Shining” by letting their servers render your clothes and skin. Which sounds plauseable, but their network is so bogged down and lagged you’re still going relying THEM anyway to render your avatar regardless how it’s done. Granted when they annouced this they did say they were going to upgrade thier servers, but they also did say they were going to close down one of thier co-locations and consolidate everything on the other two they have.
Now how is that an upgrade? If anything that sounds like a cost cutting move dropping one of their co-locations, and then add yet more traffic load onto the remaining two co-locations. That might make things worse! They did say they were going upgrade the hardware at those two co-locations, and I assume with the removal of the 3rd, there will be less traffic going across the country on LL’s private network (VPN) which will also save costs too but hopefully latency if it can handle that much traffic (since they barely can now).
So it’s a bit of a mixed bag here. In the past, every time they have done a significant hardware upgrade, it STILL isn’t enough to handle the load. Even now with fewer users playing Second Life, it’s still not performing well. After all any region with over 25 avatars in it will start to bog down, lag, and eventually crash once it’s hit it’s “main agents” limit. Honestly i think it’s just time for them to just re-think thier networking protocols to something more efficient, optimized and streamlined. Rather than this hacked together crap that they have been building for the past ten years.