Maestro Linden We're totally satisfied with this. It would be unreasonable to expect a years-long issue to be discovered and fixed overnight. I greatly appreciate Kyle meeting with us and extending the invitation to send him future data. That's all one can really expect at this point : )

Thanks for confirming.

Hecklezz Resident That is a very good question hecklezz. I give kudos to Kyle Linden today for coming and running an official-grade test with three users today to check that very thing, We took turn rezzing various items... and as is always the case when the mechanic checks the car... "the noise disappeared". However we got some very good hints. While nothing took an extremely long time to rez (in fact, it all rezzed rather well), it did rez at different speeds for all four of us (as would be expected).

However something was different today... and we didn't have all day to run tests. The region we tested on was running at unusually good capacity, no problems no lag. Quite often that region has some performance issues, which Kyle has graciously invited me to report directly to him any time that happens. We didn't have time to run the same tests in other regions. It was mentioned among the four of us however, that this MAY happen more often on sims that are performing poorly that day, so that particular thing is going to be what we'll be looking in to.

What was established among the three users is that there are many times we have experienced mesh taking a very long time to rez. Kyle accepted this may be an issue, but we all agreed that more data is needed. That's what we'll be pursuing over the upcoming weeks. Kyle was very friendly and patient in helping us run and observe test results. As I told him: we don't expect instant fixes. Linden Lab just helping us in observing basic testing was all we could ask. I really appreciate Kyle taking his time to do so.

(As a note, I asked Kyle to look at the region background information to see how it was performing. He stated it had restarted just 8 hours prior and was at peak performance, no lag, no issues, low script count, no "drag" components. So we have discussed possibilities that could cause such a sim to have poor performance, and will be watching those meters as well.)

That's the latest up-to-date information as of today, 26-04-24.

Maestro Linden Hi Maestro. With respect, I'm not speaking of testing specifically-designed models that you already know are prime and flawless. I'm speaking as an everyday SL resident who rezzes things like Teegle horses, dragons, homes, tools, fauna, statues, etc etc... like everyone else on Second Life. When these things are rezzed they remain invisible for quite some time... often 2-3 minutes before rezzing.

I'm not the only one who has noticed this. Etheria Parrott, the creator of Dinkies, upvoted this same issue in the Viewer Feedback area... until it because apparent that the Lindens there had no interest in checking the problem out... even when I offered to help them do so (as I have in the past with texture testing).

This boils down to user experience on the SL grid, which should be the bottom line in any such report. I'll state the same thing I've stated in the past: I use the Firestorm Viewer (as do thousands of other people), and I'm not being paid to do LL's debugging work for them. I report the issue. You have at least gone to the trouble of testing this under ideal conditions using the SL Viewer... but have you tested under using one of the most popular viewers on SL? These things need to be tested both under the SL Viewer and under Firestorm-- because both viewers equally affect the SL experience.

Maestro Linden Follow up... I noticed you rezzing several identical items. It is well-known that once a mesh rezzes-- just like textures and sounds, additional rezzing is instantaneous. The sim already has everything buffered, so there is nothing to fetch from asset servers (or perish forbid, archives?). So rezzing 5 or 6 ducks in a row really isn't surprising. The only one that "counts" is the first one. But that non-textured white duck you rezzed I would have to say is a very simple, basic "no difficulties" mesh. And... I would have to wonder about your access to Second Life compared to the access of the average user. When I try to rez a mesh... or others try to rez a mesh, and it takes it a full 2 minutes to appear-- there is obviously a significant bottleneck somewhere.

If the bottleneck is Second Life itself, it needs to be tracked down and fixed. If it's in many, many user's computers, it would be nice to get with someone like me and try to figure out what that issue is so that information can be passed on to the population in general.

I will say this: I have been on SL for 22 years... and rezzing items has always been a problem. Always. With prims it's not so bad; the system seems to rez standard prims pretty quickly. But with sculpties and meshes it's an entirely different story. So SL side or user-side... enough users have problems with this that this really needs a deep-dive. And as you know from the past, I'm always ready to help you and Dan test things out-- if the test is reasonable, logical and predictably productive.

So instead of having me spend my L$ uploading a mesh file (as low as that fee is in this case), why not just hand me a duck and let me try to rez it. Why should I have to go to the trouble and fee of uploading it myself? That seems a fairly simple test. But I really don't think it's going to tell us much. Or maybe it will. Hard to say.

Eren Padar: Okay, if this bug is specific to _certain_ meshes, can you please indicate where they can be obtained? Alternatively, if you'd like to test with the duck model I suggested, you can upload it for free on the beta grid.

Another avenue of benchmarking mesh loading which might interest you is the MeshTest2 region on the beta grid, which contains a bit over 5000 mesh objects.

A typical benchmark run involves clearing the viewer cache, then logging into the region at secondlife://Aditi/secondlife/MeshTest2/5/132/87 with draw distance cranked up to 512m or so, and waiting to see how long it takes for the viewer to fetch all the meshes. You can monitor the status of mesh downloads in either the texture console ( https://wiki.secondlife.com/wiki/Texture_Console ) or by waiting for the viewer to log 'Mesh Download Quiescent', which indicates it has finished loading all the meshes in the scene:

Testing a few times on a 2020-era Mac laptop with the Second Life Release 26.1.1.23806384790 viewer, it takes the viewer about 65 seconds to finish all the mesh downloads from the time I click the 'Log In' button. On a newer Linux desktop, it takes Firestorm 7.2.3 (80036) about 45 seconds to load the meshes.

Maestro Linden Maestro, with respect I have mentioned this is a grid-wide issue, involving hundreds if not thousands of meshes. A "simplified duck test" is not going to establish baseline parameters. Running a predictably pointless rez test on one specific duck model is not going to establish a . What WILL establish a valuable data set is you and I getting together on my region, me calling in 3 or 4 friends, rezzing several random mesh items and each of us reporting how long it takes to rez in their viewer.

This is trouble-shooting 101... the kind I did for 35 years in real life. This is the method I used to discover the serious texture bottlenecks in both Firestorm and Second Life Viewers: gathering dual-validated data and turning that data over to Oren of Kitely so he could nail down the causes. He examined that data, established there was no doubt of texture bottlenecks, found those bottlenecks-- and increased texture loading speed by 15x to 30x prior speeds. He handed those fixes to Firestorm dev Bec, who implemented them immediately.

I do know how to test and troubleshoot, Maestro. I appreciate you replying here, but Linden Lab needs to be willing to take the time to troubleshoot this properly. If your "solution" is to have your CUSTOMERS do all the work... then let's just drop it and let SL continue to malfunction as it is now.

I already have some ideas on what's causing this problem, but I'm not going to pursue those tests if LL isn't willing to conduct the tests that are initially necessary. It's not my grid, not my reputation, not my problem. Squarely stated: this is LL's problem and responsibility to solve. If you believe I'm the only user experiencing this issue, that would be incorrect.