Do sleeping servers wake-up fast enough?

[Yordie Sands]

If I'm boating on the Blake Sea and I approach a sim that is sleeping, will that sim wake-up fast enough for me to cross without incident?

In the past, when threaded region crossing were working well, I was able to speed across an unlimited number of sims (once I did over 100) without incident. Now, it is sporadic. At times, I've crashed on the first sim crossing and the farthest I've gone was 19 sims one evening before crashing. Crashes are the type where you loose control of your avi as they sail off across non-existent sims.

I don't remember the name of the server technology that places servers asleep when not being used, but I've read that this has been implemented. Could the sleeping servers account for the crashes I've experienced?

[Theresa Tennyson]

As soon as you enter one sim all the adjacent sims spool up to full speed automatically before you arrive in them. The "wake-up" is instant - they are actually always running, they just run at a slower simulator frame rate.

[Yordie Sands]

Hi Theresa... I hoped that's how it works. Any idea why visually and on the mini-map, sims appear to be down? It is very puzzling what is happening. Also, what is the terminology for the sleep server technoloy? Thank you, Yordie

[Theresa Tennyson]

---

Yordie Sands wrote:

Hi Theresa... I hoped that's how it works. Any idea why visually and on the mini-map, sims appear to be down? It is very puzzling what is happening. Also, what is the terminology for the sleep server technoloy? Thank you, Yordie

---

The sleep technology is called "Region Idling" and it's been pretty trouble-free. Today you'll see sims going down and back up all day on the map because they're running the weekly server restarts (which haven't been done for a few weeks do to the Holitays). Also there are periodic map glitches where the "tiles" (graphic cartoons of the regions) showing up occasionally, often causing long blank strips through continents. The regions themselves should be fine.

[Malate Banufong]

Hello Yordie

Here is a bug that a fix for is being rolled out this week on RC Magnum -

Fixed a bug on region crossing that would cause the viewer to cease getting updates from the recently departed region and would cause the region to "disappear" momentarily about 50 seconds after departure (BUG-141)

Also for BlueSteel and LeTigre this is being rolled out this week -

This is the region crossing improvement project

• Scheduled 2013-01-09, for RC BlueSteel and RC LeTigre
• Contains
  • Added background serialization of the Agent and their attachments on region crossing.  Reduces the burden of region crossing on sim performance.

There seems to be a lot of work going on to fix sim crossing problems with vehicles so we will see how things develop this week. The Blake has a lot of RC sims so it's a good place to experience any improvements.

I had noticed that sim crossing has been less good over Christmas and the New Year period at times, experiencing the same issues that you have, but it was OK before and I expect it will be OK again soon as LL are back at work.

Regards, Mal

[WolfBaginski Bearsfoot]

I think Yordie is referring to the situation when a sim doesn't appear to exist on the mini-map, as you close on the border. What I too often see is that Sim A is visible, and I'm about to cross into Sim B, also visible, but the next sim in the chain, Sim C, never appears in the mini-map. It seems very likely, from my experience, that it's the A-B crossing that is the root of all that happens thereafter, but if my draw distince is set to reveal Sim C before I leave Sim A, and it doesn't appear, it is pretty well certain that the A-B crossing is going to fail.

 

I tested this with aircraft at altitude so that most of the slow-to-download scenery is out of draw distance.

 

Like Yordie, I have also seen a deterioration in sim-crossing performance, particularly since Pathfinding was rleased grid-wide. I've seen it claimed that the roll-out of Pathfinding messed up any relationship between physical server location and grid location, which slowed the necessary network traffic, but I cannot rule out other elements of the problem being Pathfinding related.

 

The Lindens, of course, don't talk much. It's very easy to think that the Viewer code has accumulated a huge amount of crud over the years, patches upon patches, just as any other software does. The Viewer got a clean start with Viewer 2, but little snippets I hear suggests that the Server has never had that,

One account I was reading over Christmas suggested that no real-world programming project is well enough designed to satisfy a Computer Science Professor. Too much of the final product is hasty patching.

 

I'm sceptical about the prospects for improvement, without some major changes in how things are done. Clean start code? Does Linden Labs still have the resources to do that? Can it afford not to do that?

 

[Yordie Sands]

Hi Theresa... sounds like the "region idling" isn't the cause of problems.  Thanks for your explanations.

[Yordie Sands]

Very interesting info, Mal! Thank you. I read that they were retesting threaded region crossings in the past couple weeks.

As you mention, we saw sporatic results on sim crossing during that period. We generally boat down from SL New England to the Black Sea, about 20 sims and sometimes we had very wierd results. I hope the fixes work. I love boating in the Blake Sea and the eastern regions surrounding it.

Kind regards, Yordie.

[Yordie Sands]

Your explanation is much better than mine, Wolf. I have indeed experienced the type of problems you've encountered. Thanks for adding the clarity.