Soon Second Life will introduce region idling, a new “feature” intended to make SL servers run more efficiently. The intended way it’s supposed to work is if that there is no one on the region, and no one is looking into the region, that the region would go into a sort of “low power” mode. Before even I get to see it in operation, I already have several problems with this.
In the past Linden Lab has vehemently sworn up and down that regions on some piece of hardware do not affect other regions utilizing that same hardware. So obviously this must be a power related cost saving measure, and not the fact that regions actually do impact other regions performance. For the record I’ve filed several JIRAs reporting poor performance, for no discernible reason other than sharing hardware with a resource hog of a region. But for now, let’s pretend that’s not the case. Let’s pretend LL is doing this to save money on their power bills.
Here’s the part where I get extra irate. Rather than investing money Region Idling, you should be investing money in some VERY SERIOUS JIRAs that eat up a TON of your customer support resources. At this point having people on customer support must cost more than a region being online does because you’ve shed thousands of regions in the past year. That’s how you’re going to save a ton of money. Make estates easy and manageable to the point where estate owners don’t need to contact support.
As a customer who pays SEVERAL THOUSANDS of dollars to Linden Lab PER MONTH for the service, and as an estate owner I should be able to decide what I do with my regions. As it stands now, this isn’t what I signed up for. I sure don’t have things running full time, but if I want my region to be full automation even while people aren’t on it, that’s what I signed up for. I should be allowed to script up a fully automated eco-system and check in on it’s progress when I feel like it. Not to only see that it’s gone only 1/10th of it’s intended pace. As it stands now the price for a full region isn’t worth it anymore. I’ve seen nothing but value of service decline over the past 5 years. Currently homesteads are limited to 20 avs per region, live support only asks you questions until the problem goes away on it’s own or you leave in utter frustration, and concierge phone support isn’t even 24-7 anymore. On top of all that we’re getting the performance equivalent of a virtual host that is more costly than dedicated host pricing. Chew on that.
So now that region idling is here, what’s the trade off for us? Do estate owners who pay the bulk of supporting the grid get to have their regions on full time? Are all regions capable of holding more avatars now? What kind of added features are we going to see out of this? Is this just a crap trade off for pathfinding? Are our tier rates going to finally be lowered? Time will only tell. There’s one thing I can say to any Linden reading this. I am really fed up with the platforms recent choices — make a favorable change soon.
What an intense month for the Second Life grid! Marketplace, search, hardware failure, and general asset server issues made a lot of people upset. However we’ve seemed to pull up out of that nose dive only in the last few days of April. Things seem stable (as they get) for now. Here’s this months list of important things!
Important Server Fixes
- [WEB-4678] Regions not showing in inworld search. This JIRA is just the tip of the iceburg for many search related problems. I myself, historically, don’t use search that often. But this month I’ve been using search to see how my available parcels show up. Most times they do, but sometimes they don’t. For now there seems to be no rhyme or reason as to why search intermittently fails.
- [SVC-7837] Filtering by object name or owner name in top scripts/top colliders no longer works. Estates are already hard enough to manage when “estate wide” functions don’t work. Now the most very basic functions are broken. Rather than try to fidget with broken inworld tools, I’ll just call Concierge to do it for me. :P
- [SCR-312] llSetRegionPos() does not return false when trying to move into nonexistent region. This one is a bit close to my heart, as I had hoped the llSetRegionPos() function would be smart enough to return a FALSE rather than try to cross into a non-existent region during a rolling restart. Getting a dozen or so IMs every time a region restarts because objects being returned is not cool.
Important Client Fixes
- [VWR-15729] and [MAINT-720] Unknown failure mode causes uncontrolled network flooding. These issues are closed, but they shouldn’t be. Essentially the problem is if that you use the voice feature in Second Life, after you log out or quit the provider (Vivox) can keep sending you data for over an hour. But get this, the issue was closed as “Won’t Fix”, with no reasoning added to the issue, but a comment is later added to the issue where Mercille Linden says that it’s a Vivox issue that they have no control over. Tateru Nino makes an excellent comment about how they don’t have direct control over the issue, but LL can at the very least talk to Vivox about it, and help to fix the issue. The only known fix is to DELETE SLVoice.exe, and with that lose your SL voice features. But SL voice has historically been filled with problems, perhaps it’s time for LL to find a different voice provider?
- [VWR-248] Inexplicable folding of Avatars such that they are walking around with their heads up their arses. Even though this JIRA was filed in 2007, this bug has been in SL since I started way back in 2005. This bug doesn’t happen very often, but it’s plenty humorous when it does happen. It’s just one of those annoying, but silly things in SL.
- [ALL THE JIRAS] Alpha Flicker Issues Please Linden Lab. Dedicate someone to fixing the alpha sorting issues that have been around since SL BEGAN. SL would be a much better place if my client wasn’t going to give me an epileptic fit from staring at some sculpt trees.
My Wish List
- Asset and LSL function versioning. More often than not any feature that we’d like to upgrade gets rebuffed by the lab with their catch phrase “It’ll break existing content”. I think it’s time to start versioning the assets. Mesh2, and llSensor2() are just examples. For instance Mesh2 could have all the qualities of a mesh in SL as of now, plus future upgrades, or llSensor2() could be a more fine grained control of sensing things around us, by owner, on land, etc.
Tip of the Hat
- We can finally fly to 4000+ meters in the sky without a flight assist. This is another one of those long desired, but simple user experience features that have been years in the making.