435 Havok4 Early Adopter regions updated with 16 fixes 2008-02-07

Quick Status

Havok4 Beta Server v1.19.0.79349 just completed testing on the Beta Preview, and has now been deployed to 435 regions in the Early Adopter Program. Thank you for helping to test the new Second Life simulator. When we picked an initial target for the Early Adopter Program I thought “well ok – 200 seems achievable”, then after getting poked a few times to pick a higher target, I got comfortable with the idea that maybe we would get interest to convert 300 regions. Little did I know that our biggest challenge would be keeping up with the expanding demand!

We have been making weekly progress on high priority issues, and with this build have removed some key issues that have been concerns in the first Early Adopter test phase. Please see the next section for details on the changes in this release.

What Has Changed In This Version?

The new build (Havok4 Beta Server v1.19.0.79349) has been deployed to both the Beta Preview and Second Life Early Adopters. Changes in this build include (SVC items were submitted on the public jira issue tracker / DEV items are internally discovered issues):

DEV-9324: Avatars no longer fall through prim stages and floors with many avatars on a region and high physics loadDEV-9236: Altitude is now held consistently with flight assist at high altitudes

SVC-1089: Avatar inertia now maintained when jumping (fix basketball and other sports)

SVC-1073: llGetStartParameter() now returns real value, and not always zero

SVC-1146 and SVC-38: LSL scripting – llSetPrimitiveParams now behaves correctly when setting options on different sides. The solution implemented may affect existing scripts that rely on work-arounds for bugs in the old, pre-Havok4 behavior and we are looking for feedback on those cases. Please comment in those issues with specific parameter orders that you need but no longer work and include the needed results.

SVC-1301: Builds packaged with Rez-Foo now build correctly

DEV-5014: Spheres using “Dimple Begin” now collide correctly

DEV-8718: LSL scripting – llTargetOmega objects now spin correctly without being selected

This is “take 2 on DEV-8718″: In some cases child objects would still not spin correctly without being selected after the last fix. This time it is believed that child objects will also behave. Please let us know if this is not the case!

DEV-9240: Create copy selected now generates gapless alignment to the adjacent object

DEV-9329: A few regions experienced physical objects being thrown off-world at startup with the beta simulator. Physical objects with certain internal problems will no longer be thrown off-world at startup.

DEV-9416: Increased distance that objects can be under the terrain before being thrown off-world

DEV-9643: Mega prim objects now link as they did on Second Life (some mega prim linkages were not working on the Havok4-based simulator that did work on the main Second Life system)

DEV-9774: llTargetOmega objects no longer stop spinning after you select them

DEV-9836: Avatar falling velocities are now correct

DEV-10003: Avatar now moves when pushed by attachment when on terrain

DEV-10015: Simulator no longer crashes when ray trace extends past edge of world

What Does It Mean To Be Part Of A Beta Program?

Please note that the Early Adopter Program is a large scale beta test of a new version of the Second Life simulator (server program). When you opt in, you will be using software that is new, and likely still has some rough edges. The good news is that with each release we have resolved a set of important issues. At the same time it is important to remember that the point of this process is to find the problems that we cannot find until we’re running “for real” with real avatars and content.

That means that there is some risk of problems, a risk that we have tried to minimize, but cannot entirely eliminate (if we could, we wouldn’t need to test!). What you get as a benefit of being in the program is the opportunity to try out the new simulator early, help to find the remaining issues, and be directly involved with helping to be sure that the new Second Life simulator will work well supporting your regions before it is rolled out to the whole Second Life Grid.

How Do I Join The Early Adopter Program?

If you own a private island and would like to have your region converted to the beta Havok4-based simulator, please follow the instructions in the program announcement post. (Please do not add a comment here saying “add me!” and expect that your region will be converted just by this post – we will only convert regions based on tickets submitted by the estate owner.)

At the moment a bit over half of the 300+ regions are deployed, and we will update the rest of them by mid-day tomorrow.

Am I In A Havok4 Beta Simulator Region?

You can tell whether you are in a region that is supported by the standard Second Life simulator or the Havok4 Beta simulator by opening Help > About Second Life… in your viewer. The fourth text line will start with one of two strings:

* If the fourth text line starts with “Second Life Server”, then you are in a region supported by the standard Second Life simulator

* If the fourth text line starts with “Havok4 Beta Server”, then you are in a region supported by the Havok4 Beta simulator

How Do I Report A Problem?

To submit a Havok4 bug:
1) Log in to the Public Issue Tracker at https://jira.secondlife.com/ page with your Second Life avatar name and password.
2) Click “Create New Issue”
3) Set PROJECT to “2 Second Life Service – SVC” and ISSUE TYPE to “Bug”
4) On the next page, type step-by-step instructions for how to create the problem, and if specific products are involved include the product vendor name and specific product name and version.
5) VERY IMPORTANT: Please set COMPONENTS = “Physics” and AFFECTS VERSIONS = “Havok4 Beta”. If you do not choose these settings, the Havok4 team may not see your bug report for a while!

For a demo of the jira public issue tracker, see Torley’s great video “how to report a bug“. Again, please be sure to set COMPONENTS = “Physics” and AFFECTS VERSIONS = “Havok4 Beta”!

Best regards,

Sidewinder Linden
Havok4 Program Manager

P.S. Please keep your posts on-topic (just had to repeat this – everyone’s doing a great job at keeping comments on these posts high in content and on target).

This entry was posted in Announcements & News, Bugs & Fixes, Preview Grid. Bookmark the permalink.

83 Responses to 435 Havok4 Early Adopter regions updated with 16 fixes 2008-02-07

  1. Lorrie says:

    ok seriosly now why bother updating post like this when half the SL residents cant get online why not focus on gettin logins working then maybe add this but i think this is a bit out of place since no one has bothered to update the logins blig yet u up date other. its time to priorotise.

  2. Thank you, Sidewinder. Updates on the H4 progress are a beautiful thing.
    ;D

  3. Elvis Orbit says:

    Great thanks Sidewider ! Look forward to taking a look when I get home from work on my SIM.

    @1
    This is a Havok 4 team post witch is a separate work group from those dealing with the Login problems. And I was waiting for this post all day so it is a welcome post.

  4. Chaos Markstein says:

    Lorrie:

    This is them trying to fix logins, sometimes the reason you cannot log in or experience problems is due to heavy load on the servers. Havok1 is a laggy and overbloated system, where as Havok4.2 (the new SL Havok beta system) is going to take ALLOT of load off of the servers, meaning they have chance 2 do other stuff such as logging in quicker and loading textures a bit quicker.

    Sometimes you have to look away from the problem 2 find the problem

  5. Desmond Shang says:

    This is great stuff Sidewinder, keep it rolling!

    We’ll keep pounding on the regions as best we can.

  6. Mako Minogue says:

    Excellent news, kudos to the entire H4 team for the bug fixes and rollout work.

    We have been beating the tar out of the H4 code in my sim and so far it’s been very impressive in spite of the bugs.

    Good stuff!

    Mako

  7. Vittorio Beerbaum says:

    Anyway to figure out via SLS if you’re on H1 or H2? This would be a very important features for those of us that are developing phsx tools, especially during this startup, since there you would encounter different engines that may requires slight different setups (parms tweaking).

  8. Digital Digital says:

    Awesome Love the updates keep em coming :))

  9. Pingback: MMODump.com » 435 Havok4 Early Adopter regions updated with 16 fixes 2008-02-07

  10. Margo Oliva says:

    where can we go to get help for the login situation the secondlife website is not loading?

  11. Kimo Junot says:

    I really think it is great and all that you guys are trying to make SL a better place and what not but more and more this is starting to remind me of a ford pinto I owned back in the 70’s. It seemed like no matter what I did to it…endless tune ups…swapping engines..new transmissions..etc.. it still ran terrible O_o

  12. Harvey says:

    Great work Sidewinder and team 🙂

    @11 if you can’t get to the website at all it’s probably to do with your isp and is probably the reason you can’t login. If you can get to the website but it doesnt work as expected then it’s something at LL usually.

    You might be able to get an email through that will be read by the support team perhaps support@lindenlab.com not sure if that will work though

  13. Pingback: Massively

  14. Sidewinder Linden says:

    @7 Vittorio: The topic of recognizing which Havok version you are on from LSL has come up a few times in the last month. At office hours recently we talked about it, and have spent a while discussing options internally. We ended up deciding not to do anything about this – not because it isn’t a good idea, but because of “right answers” and timing. There are a lot of things about the simulator that LSL (or Mono) scripts might like to know about, and we thought that providing only information about Havok version would be limiting and then likely a piece of implementation that we’d want to replace in fairly short order – with a more generic solution.

    We didn’t think we could design the “right” long term simulator information query solution, implement it, and get it tested and through the pipeline quickly enough that product vendors and scripters would:
    a) find out about it
    b) have time to implement it
    c) have time to distribute product updates and have them take hold
    before
    d) the new Havok4-based simulator was finalized and rolled out to the whole Second LIfe system – at which point being able to sense the Havok version becomes moot.

    Regards,

    Sidewinder

  15. Margo Oliva says:

    On the login screen it just freezes at veryfying protocol version, then it gives a message that says unable to connect to second life. The system may be down, Please try again in a few mins,or click help for advice and link to the system status webpage. But when I go to the webpage none of the likes work there all dead links. I know its not my computer because I had the HP people and the nortons anti virus people working on it this morning and neither one of them could get me logged in. What is going on?

  16. James Benedek says:

    YES HAVOK 4 IS AMAZING 😀 my estate owner adopted the sim i rent early 😛 sooo fun. I love how progressive it has become.

    Good Job Sidewinder, Dan and Team.

  17. Wyald Woolley says:

    I have found Havok4 to work for me rather well. A bit laggy, but that was all I was aware of…if I didn’t know I was on a H4 sim, there was no strangeness to tip me off. Although I know that some actions and vehicles still need tweaking. Once H4 is tweaked a bit more I believe it will be a good strong step forward for SL. I applaud any changes that update the underlying code to SL to bring on more stability and less lag…without adding unnecessary shiney.

    I also find it amazing that you can make such fundamental physics changes from H1 to H4 and have it still be such a smooth-seeming change to me…congratulations!

    I once had a link to the list of open issues, Is such a list still available?

  18. Pingback: Second Life Sucks

  19. Bob Wellman says:

    As there are 430 Havok4 sims running on the main grid is there any chance you could publish a list of them on the SL website somewhere so interested parties could go to one and try it out.

  20. Sean Heying says:

    I picked up 300 objects in my Havok4 sim as a single conglomerate. There was also one locked object in that conglomerate. When I rezzed it at a new altitude the items just deleted requiring a rollback.

    I am currently trying to test why, and how to duplicate so I can resubmit… but maybe watch your rezzing if you do similar things.

    It may not be Havok, but due to the high loss of content I thought I would say in warning just in case.

  21. Lindsey Warwick says:

    Why cant they make the physics work as awesome as SOURCE!!

  22. Muziekfreak1980 Miles says:

    Im not into all this technically, just a builder. I do understand all this tho. In farmers langue, havok is like de-stressing the overloaded servers at the moment. WELL GOOD! HANDSUP FOR THE LINDEN..

    PS not the right topic, but why is windlight still so unstable from time to time?

  23. @21: This is why it is a BETA

    They have been working to make the physics work as good (or rather, the same as) the Havok1 sims.

  24. Vittorio Beerbaum says:

    @ Sidewinder : makes sense, i just didn’t known about the estimated time of spreading H4 gridwide, i thought it would have been so slowly (read: many months with the number of H4 simulatros increasing time to time) to motivate a LSL function implementation (even if it wouldn’t been utilized after then). If you say that a such developing (of function) would require more time than adopting H4 everywhere, then there’s (obviously) no sense to made it available.

    @ 22: i’ve asked myself the same question.. i’ve come to my conclusion with the latest WL release… it looks like (or it would effectively be) a huge step behind because of a reason: they are merging both (windlight and “normal” client) into a single client (because the plan, i guess, is to make a single client.. not to have WL in beta stage forever…). So we’re loosing all the “benefits” of WL (gfx apart… indeed): it’s more unstable, it does have a hugly GUI (buttons placed in apparently no-sens order…), i’m having lag and crash when i didn’t experienced a crash for months in the recent past (except those coming from the “teleport bug”). Yes, the most recent WL versions (especially the latest one) are disappointing me too.

  25. Ryu Darragh says:

    #1 Lorrie, the two issues are seperate issues and complaining about login failures on a post about H4 has to be a last ditch effort to get through. I am telling you, and anyone else reading this (as this is a common issue), Linden Labs and your PC connect through the internet. That means there are *THREE* places for failures to occur. The middle, the internet part, is not a single simple pipe between you and LL. Different kinds of traffic can go different ways and not all of Linden Labs “component parts” are in the same place. The blog can be in Nevada and the website in Washington and the servers in Texas, California and Boston.. in otherwords, a clog in the pipe between *you* and the login servers can keep you from getting online and *still* let you access the web site or the blog with ease ! Get with it people. The internet is an unlikely confabulation of calamitous clanking nuts and bolts, twine and baling wire and it’s a miracle it works at all 😛

  26. Ryu Darragh says:

    #20, Sean, this is an issue with H1 as well. When you “coalesce” objects into your inventory, they can suffer the broken perms contamination issue. Be sure all objects are unlocked and permed properly before linking or coalescing. And being “locked” is one of those. This happened to a build of approximately 320 prims I had made that had some locked parts. It all became nomod and locked when rezzed. This is an asset server and permissions issue. Another important point is that although a rezzer (I made one based on WarpPos) can send objects to 4096m on an H4 region, if you try to reposition manually or if the prims are locked above the clients limit of 768m, the prims will either self delete or jump to 768m. Check there for a rather flat looking build 😛

  27. Pamagester Darracq says:

    WOW thanks lindenlabs! you have fixed pretty much all the havok 4 issues that had been bothering us (the people that have land in early adopter sims) for now. Well we might still find a lot of bugs but now that not being able to fly with assistants, walking across prims where few and the issue with the rezers that used the prams well the sims feel like back to normal but with added features YAY.
    THANK YOU!

  28. Mike1 says:

    So do we have to re-down load or what! I still can’t get on?

  29. U M says:

    For the sake of not having my post removed. How stable are the teleports with h4? FRom what i noticing when going from H4 to another class server teleports die with avies in them. Or just outright fail………. Stabilty is indeed important but so is teleporting. If people can`t get out of those H4 island there indeed needs more testing.

  30. elvis orbit says:

    @29
    I have Havok 4 on one of my sims and have never had problems teleporting to and from H1 sims.

  31. Captain Noarlunga says:

    Great progress with Havok 4 and definite improvements with WindLight; although as both are still Beta, there will be more iterations before general release. The one thing that bothers me though, is that despite these items providing improved performance, there are still major underlying issues that need to be resolved in the basic system….. e.g. Crashing on TP ….. which is very prevalent, to the point of almost having to relog every time you TP. I just wish this would be resolved, because if it isnt, it doesnt matter a damn how good Havok and WindLight are !

  32. Captain Noarlunga says:

    Addendum to @31….

    and it isnt my PC…I get the same problems on 3 different PC’s and with all Clients (Normal, RC & WindLight) and on 2 different ISP’s. The problem is definately Server Side.

  33. Ioji Kutanaga says:

    Any update is a good thing. This is a massive community, and i would compare it to repairing the new york city road system. things need to be done one at a time, because if it is done all at once, traffic stops entirely. Thats just how it will end up. So if havok4 takes some time, this is likely the only reason why. things like this take a lot of time and effort. personally, i am happy with everything that has happened.

  34. Neural says:

    Don’t know if this has been mentioned before, but is worth repeating, since the Lindens have a tendency to re-introduce old bugs in order to get new features/fixes out.

    This is in regards specifically to the llTargetOmega() problem that was fixed in this version.
    Should it come up again, a fix that worked in this last version was to drop in a script that had floating text (set to 0 alpha) that updated on a timer once every second. This forced a refresh from server to client, and worked like a charm

  35. Isablan Neva says:

    I cannot tell you how thrilled I am to be able to fly normally at high altitudes again. I rarely leave the sky and the past couple weeks of moving between work platforms and various projects has been a hilarious comedy routine of fly up, dive for platform, miss, fly up again, dive again, miss, fly up again, dive again, finally land.

    YAY!!!

  36. Crucial Armitage says:

    H4 is awesome I have it on one of my regions and I find my over all experience to be better. Keep up the great work Sidewinder and the rest of the H4 team!

  37. Kelly Linden says:

    I don’t think it is a good idea to post a list of the regions with havok4 – it would come across to some as an invitation to test physics on those regions, which would essentially grief them. We usually put up a havok4 sandbox region or two, I’ll see if I can find their exact names.

    TargetOmega: I actually made a couple more tweaks to catch some very edge cases, those tweaks will go out in the next update. TargetOmega is tricky, as far as the server is concerned it always works – what fails is sending the updates to all the viewers in specific cases. The reason setting the floating text works is it forces a full update to be sent to every one who can see the object. This is actually quite a waste in bandwidth because target omega data is usually a part of the much smaller ‘terse’ update. But you are correct – if it is failing to update one workaround is to do something that requires a full update like you describe.

    Source physics: Source physics has the luxury of running on your PC. It only has to deal with you and your actions and it has no network latency. For multiplayer every player is simulating physics on their own PCs and occasionally syncing up data. Additionally the vast majority of a scene in a Source game is static, it never changes. When you *know* something can’t change the physics engine can be much more efficient. SL is very different – all physics for a region is actually done server side by the simulator and the entire world could change at any time. This means everything, even walls and floors, are more dynamic and take more physics time than if we knew they weren’t going to change. We have done some thinking about “static” content in the havok4 sense, things that really aren’t going to change. We’ve thought about both allowing content creators to flag their floors and walls as unchanging and automatically detecting items that haven’t changed for a while. Before we can implement any changes along those lines we need to do some benchmarking on the performance cost of changing the ‘static’ness of objects and some measurement of how much of a performance boost we actually get. It isn’t clear if that is something we will look at for performance reasons before or after the initial havok4 release.

  38. Sidewinder Linden says:

    @29 U M: There have been teleport problems in general over the last week or so in various locations, and these are unrelated to Havok4. /Sidewinder

  39. Sidewinder Linden says:

    There are two Havok4 sandboxes on Second Life. These are *purposefully* not named Crash Me by the way… Please let folks do normal sandbox/building experiments here rather than terrorizing them with attempted region kills, ok? 🙂

    Havok4 Sandbox Dmg
    Havok4 Sandbox NoDmg

    Have at… /Sidewinder

    P.S. NoDmg is offline for some reason at the moment, and I’ll work on getting that back up as soon as possible.

  40. Sidewinder Linden says:

    @31 Captain: Crash on TP is an issue that’s being worked on by another team, and is not consistent in all regions nor at all times.. It is also not Havok4-related, so please log tickets in the trouble ticketing system on this topic rather than taking up additional comment slots on this blog entry. Thanks! /Sidewinder

  41. Sidewinder Linden says:

    Havok4 Sandbox NoDmg is now back online. /Sidewinder

  42. Sean Heying says:

    @26 Ryu, I thought a few things, like the build being at a different height, but this was shifting items from 200M to 600M, well within the current 768M limits the Havok4 beta is running (wasnt this supposed to be 1024M Sidewinder?)

    In this case the items didn’t get wrong perms, they poofed. Dissapeared from the Sim’s prim count, no warning, no return, just gone. I had to roll back to regain them all.

    There was a bug there I dont know if it’s H4 or asset or sims in general… or my silliness for wanting to move an entire skybox with rubberband select. 😉

    It could even be coalescing more than 255 objects, is it even allowed? If not, should the H4 sim warn you?

    I should try to get LLs approval to do two roll backs, one in each of my sims H4 and H1 to test this skybox move in both. When I’m more awake I might ask

  43. Sidewinder Linden says:

    @42 Sean: The ceiling will actually be raised to 4096, but there is a viewer-side check on this so it won’t be exposed until a new viewer build is done. We’d talked about this in a team meeting recently and thought we’d build a beta viewer that includes the 4096 ceiling, so that folks can proof the behavior on the Beta Preview. Timing has not yet been figured out for when exactly that modification would show up in the release viewers for Second Life.

    Per your question about moving a whole skybox… “I don’t know, but I’ll check around to see what the possible issue is there”.

    /Sidewinder

  44. Sidewinder Linden says:

    @43 Sean: I think that the skybox issue may be a coalesce issue that is likely unrelated to Havok4, as mentioned by Ryu. /Sidewinder

  45. Chryssalyn Donnelly says:

    @Sidewinder: I love how quickly you guys are working to resolve all the bugs related to the beta sims (I’m an estate manager/builder for one). Sadly, I need to report that once again DEV-9240 is not fixed. I am still getting gaps/misalignment when copying off the face of a prim. Seems now to happen most often with prim that are not at an increment of 90 degrees on rotation.

  46. Chryssalyn Donnelly says:

    Well, scratch that, it’s happening even at increments of 90 degrees. Example: 5×5 prim with x and y values of 174.75 and 167.5, respectively. Copied prim comes out with x and y values of 174.751 and 167.45, respectively.

  47. Sidewinder Linden says:

    @45 Chryssalyn: No problem – and the *good* news about finding something new *now* is that far more people won’t find it broken later, because we’ll be able to fix it before we roll it to the rest of Second Life. Keep on finding the edge cases, and be happy about it – it helps! Would you be willing to try some more permutations, and send me a notecard in world with specific cases that seem not to work, with a recipe for how we can step-by-step reproduce them? Thanks, Sidewinder

  48. Chryssalyn Donnelly says:

    Sure thing 🙂 I’m in the middle of a big build right now, so as I find them I’ll just jot them down.

  49. I seem to be having an Inventory, AV, HUD-Script ~ Rez problem with 1.19.0 (79185)
    If I load Inventory, AV, HUD-Scripts in Havok4 Beta Server 1.19.0.79349 then go to the next sim that is not Havok4 Some HUDs detach, some scripts stop working and some of my AV attachments detach like body crusher for Tiny.
    Then if I want to switch AVs, I can’t detach prim body parts from AV. Seems to be an asset-rez issue. Map still takes about 1 minute to open and crash on TP too. TP from Havok4 to regular sim has the detach problem too. Before 1.19.0 (79185) things were more stable with 1.18.6 (76886) for me. I have noticed Havok4 ~ 1.19.0.79349 still having same flight and other little glitches and looking forward to the next update.

    To get any work done I have to go back to 1.18.5 (3) since 1.19.0 (79185) has become unusable
    Well keep trying, Thanks ,,,=^_^=,,,

  50. Sidewinder Linden says:

    @49 nokithecat: Could you please itemize a specific case of hud detach, with which products to use, and steps to follow to create the problem, so that we can replicate the issue and resolve it? Asset rez delays are likely to not be Havok4 issues, but if this is repeatable it will be useful information. Thanks, Sidewinder

  51. Sidewinder Linden says:

    A note regarding teleports, walking and flying between Havok1-based and Havok4-based simulators…

    There have been various comments suggesting that teleports from Havok1 to Havok4 simulators are sometimes odd, and in some cases you can’t walk back and forth or can’t fly back and forth, or in some cases that one region looks like water between two regions, but only from one side of the region boundary.

    None of these are Havok4 issues. What they usually *are* is an indication of a network problem at the Linden Lab data centers, and these occur on Havok1 as well.

    Please talk to your concierge if you see any of these symptoms, and they will have our operations group look at the hosts involved to see what is happening.

    We did discover a network routing configuration problem last week, and one this week, affecting different blocks of Havok4-enabled hosts, and I believe that this issue was the cause of all of these problem reports.

    At the moment, I believe that these issues are resolved, so if you see them again, please let concierge know.

    Thanks,

    Sidewinder

    P.S. nokithecat – I “half-suspect” that this could have also caused some of the region crossing issues that you’ve described, so if you could re-test now that we have resolved the current routing problem I’m hoping you’ll find the issue to be resolved.

  52. Storyof Oh says:

    logins STILL arent fixed…can someone plug the UK back in? old viewer and new release 1.19.1 showing DNS problems still…

  53. Sidewinder Linden says:

    @storyof: Please try a traceroute to http://www.secondlife.com (“traceroute http://www.secondlife.com” on Mac OS X in terminal or “tracert http://www.secondlife.com” on Windows). If the traceroute stops before you get to sanfranciscoN.level3.com, then it is definitely an ISP routing issue and not something we have any control nor influence over. DNS is likewise to be an issue closer to you. (I am across the US and not seeing these problems, for what it’s worth). Regards, Sidewinder

  54. Shadowquine says:

    @20 & 26

    From what I’ve gathered, I haven’t been able to rez any locked object from my inventory in an H4 sim. I can rez the objects normally in an H1 sim. One object in particular is a roughly 1000 prim backup of a building with a few various pieces of it that I never unlocked before backing up in an emergency. Never once had an issue rezzing it ever since I took the copy back in 2005, but now I can’t rez the building on H4. I wonder if the problem is related.

    I’m not sure if pointing out JIRA issues is okay in the blog but meh. I’ve had SVC-1198 in since Jan 15th, though I’m going through all of my objects that are no modify to make sure they’re not locked so I don’t lose those objects entirely when H4 is finally rolled out.

  55. U M says:

    38 Sidewinder Linden Says:

    February 8th, 2008 at 9:02 AM PST
    @29 U M: There have been teleport problems in general over the last week or so in various locations, and these are unrelated to Havok4. /Sidewinder

    Ok sidewinder…………..
    But why is it that I experiencing this when trying to Teleport out of these H4 projects? Sometimes dieing in the teleporter when going from H4 to a non H4 simulator. Now I do know from experience that this is fact. It doesnt happen in anyother situlation. Cooccurance? Your refering to? Maybe But I think you better look at the actions of teleports alittle closer because you people are missing something. These simulators eitehr have alot of campers in them. Or none at all.

  56. I been doing a multitude of retesting in the 6 Avilion sims and the asset-script issues have dramatically reduced.
    Crossing from Havoc4 to regular sims are normal now.
    I only have one or two objects/scripts on my AV and or HUD that will randomly stop working in TP-boarder crossing.
    It seems to be a network asset data transfer issue, like a time out and the full contents are not loaded or limit detaching of objects.
    Its not the same object-script that will have an issue but a random prim or script that will stop working or not detaching.
    Some HUD or scripted object ( body crusher) will work if taken back to inventory and wear again.
    The scripted & non-scripted objects that fail to work also will show contents loading & missing object name and can’t be taken to inventory.
    Flight in Havoc4 with flight assist is stable below 85m but above that elevation my AV keeps floating away.
    I would say this issue is mostly fixed for now just a little more network improvement and it might solve the issue.
    I have noticed the sims stabilize and not network flicker.
    The only other issue I have is with 1.19.0 (79185) Map & Teleport.
    If I relog, I get one quick opening of the map for the full time I’m logged in.
    All other map opening is about 1 minute on average to max of 2 minute.
    Then the TP is mostly (97-98%) ok with all my AV attachments intact if I wait.
    If I do a TP from inventory my system will freeze like it does for opening the map but after a minute it will disconnect me.

    I’m looking forward to future improvements and updates.
    The lighting and color of Windlight is a Great Success!
    Thank you very much for the progress you are attaining!
    ,,,=^_^=,,,

  57. Sidewinder Linden says:

    @55 U M: Please read post 51… /Sidewinder

  58. Sidewinder Linden says:

    @54 Shadowquine: My apologies – I missed the importance of that bug report. We will take a look at SVC-1198 this week. Is this still a problem with the version deployed yesterday? (In other words the locked prim rez is still not working correctly if attempted today? /Sidewinder

  59. U M says:

    sidewinder………..said “There have been various comments suggesting that teleports from Havok1 to Havok4 simulators are sometimes odd”

    Ok i can see some sence in this….. But what i am saying is Havok 4 to Havok 1……. From what i see from going from H1 TO H4 is not a problem. Its going from H4 to H1…….Walking backwards etc yes i am experience this. But not to the H1 to H4? something is not right here. I understand to a point what your trying to do here. But it seems to be a little confusing trying to piece togther all these diff factors and getting a series of reason why all this is occuring.

    If what you said is true. Then would all simulators be effected by this as you said
    “None of these are Havok4 issues. What they usually *are* is an indication of a network problem at the Linden Lab data centers, and these occur on Havok1 as well. ”

    And not a selected number of given simulators?

    Many of us want the H4 to work. We all been dreaming of this next step in the evo on SL. I sure don`t want it to fail. I just don`t want to see a rush work job done on this and then blowup..3 years I have waited for H4…….

    Usagi

  60. Sidewinder Linden says:

    @59 U M: If you would like to talk about routing issues offline (ie. via IM in-world) I’d be happy to talk about this in more detail, but the bottom line is that occasionally there are routing issues between either individual servers (each of which host up to 16 regions) or a rack of servers and one or more other racks. Routing issues are not always bidirectional – they can affect traffic in only one direction between a pair of hosts, and thus create problems from “host a” to “host b” and not show a problem the other direction.

    In terms of understanding the factors, we put up a couple of new racks of servers to support the Early Adopter program. These hosts were not in the normal rotation of the Second Life grid before and had a configuration problem – one that we now believe to be resolved.

    The bottom line is that the symptoms of not being able to walk, teleport or fly to another simulator, in the current architecture, is a symptom of an inability to route packets correctly from one server to another, and not a problem with the higher level programming that is the simulation engine itself. This routing problem happens with Havok1 to Havok1 traversals, and has been around for quite some time. Because routing problems at the network layer can happen in one direction – not necessarily in both at the same time – the symptoms of seeing a “barrier” between simulators can also show up from one direction rather than both.

    Regards and hope this helps to make more sense of this,

    Sidewinder

  61. Sean Heying says:

    I am also getting “hud detach” but am a little more advanced in my testing… still not enough for a jira post yet.

    For me it happens sometmes between H1 and H1 sims, sometimes on login. I see the HUD on the screen but cant detach it and some or all of the scripts dont work.

    It definately happens if I fill all my hud attach points and some of my body attach points… it often happens when I have 75% HUD attaches filled. It never happens if I have 50% HUD attach points filled.

    On some occasions even my collar does this… when the collar stops working I can see it but no one else can… it’s like it falls off. I would guess that HUDs would be the same if they could be seen by others.

    I have battled with his detaching HUD issue way longer than Havok has been in grid… some clients are better than others, some are so bad they are unusable.

  62. Sean Heying says:

    @58/54… that is my bug and I just tested on my skybox.

    If I unlocked the single prim I could take and rez in H4.

    The other day when I rezzed, as it contained thousands of $ of no copy items they just poofed. The test in SVC-1198 confirms it if you get an alt to make a nocopy prim which you then, buy, lock and take. It is deleted.

  63. Shadowquine says:

    Sidewinder,

    I’ve just been doing some testing on it. I still can’t rez locked objects in Havok 4, and I’ve also been able to recreate content loss when trying to rez no-copy object(s) that are taken with a locked and copiable object.

  64. Wanja Lubitsch says:

    On my windows, tracert does not work with the protocol indicator.
    So it should read
    tracert http://www.secondlife.com
    and not
    tracert http://www.secondlife.com

  65. Wanja Lubitsch says:

    and.. this blog adds the “http://” to the address… hehe..

    A few tries of finding the right format… since there no help on format link here somewhere

    tracert http://www.secondlife.com
    tracert http://www.secondlife.com
    “tracert http://www.secondlife.com

    …well, tracert only works without the “http://”

  66. Sidewinder Linden says:

    Sean and Shadowquine: It would be most helpful if you could check the sequence that you are using to test on Havok1 as well as Havok4, in order to best determine which team should address these issues. It sounds like some (or maybe all) of these are generic – i.e. not in Havok4 only, and it would save time to resolution if I could make sure to bring these issues to the right dev team on the first try. Thanks! /Sidewinder

  67. Sidewinder Linden says:

    @65 Wanja: Yes, this blogging engine is “too helpful” about web addresses, turning anything that looks like a URI into something with an HTTP:// on it… for tracert and traceroute, the address of the target machine should *not* include the http:// protocol specifier at the start. /Sidewinder

  68. Sidewinder Linden says:

    @61 Sean: Is there one HUD that displays the detach misbehavior more than others, or particular pairs of regions that seem to be more likely to show the region when crossing? This does sound like a generic simulator issue as opposed to specifically related to the Havok4 Beta, but if it were possible to get this closer to reproducible on demand it might be more easy to tackle… Thanks, Sidewinder

  69. Shadowquine says:

    Just did a rez test in a Havok 1 sim. All locked objects I couldn’t rez on H4 rezzed perfectly fine. I tried rezzing a no-copy object with a locked copiable prim and it worked fine. When I took it back just as I placed and tried to rez it on a Havok 4 sim, it disappeared. It looks like H1 can handle the locked setting just fine, however H4 still doesn’t work at all.

  70. Sidewinder Linden says:

    @69 Shadowquine: Thanks! I’ll put this on the docket to look at as soon as possible. Sorry for missing that case – we’ll get it fixed. /Sidewinder

  71. Sean Heying says:

    My HUD positions day to day are:

    Omicron 1.4 – Top Right
    Huddles – Top Left
    Bottom Left – Owner Online (A simple dataserver script on an Avs UUID to see when they log in)
    Bottom – Mystitool
    Bottom Right – Emdash
    RealP*n*s R.P. Hud – Centre

    The Mystitool and Emdash will often detach on login and region crossings, H4 to H1, H1 to H4 and H1 to H1, especially with the production client. Nothing yet reproducable.

    Ive had the huddle detach once in January. The RealPe*i*s has detached only once once. Owner Online has never detached. The omicron has also never detached to my memory.

    The Body attachments that will drop off are My Amathyst Collar (Chest Mounted) and my hair (Skull Mounted). I have scripted tails and ears but they never detach. My hair is not scripted and is 95 prims.

    The body attachments do not end up at my 0,0,0… They can be seen by me in their correct place when they detach so to me they look like they are still attached. If is only when I need to use them or someone mentions that I am bald do I need to relog to get them back.

    A clear visible cache cone does not cause them to restart/reattach.

    Loging out and back in causes them to attach automagically. People see me log in with my collar on when before they couldn’t see it.

    The current 19.1RC is slightly better in that the detach is not as frequent.

    I’m not conviced it’s H4.

  72. Argent Stonecutter says:

    Sidewinder: Speaking as a creator and scripter… returning the simulator version string as displayed in the About Second Life list would give us the information we need in a way that we can communicate easily with our customers and roll out updates for very quickly.

    It wouldn’t necessarily be for Havok 4 deployment, it would be a useful thing to know in any case. It would happen to be useful here, but for example if a customer comes up with a bug in a simulator we haven’t seen before and don’t have access to, then we can put in a version check and send them an update to test.

    How about it?

  73. Sean Heying says:

    My collar just detached H1 to H4 teleport.

    In H1 I got my messages: [17:13] Collar: @version=1516673
    [17:13] Collar: MoDetachWarner (NC) v1.1 active, attatched to: chest/sternum

    So at that point it was working.

    I spent about 90 minutes in that sim then TPed to my H4 sim and touched my collar to test… it was dead script wise, which means people will not see it.

    This was not a region crossing this time in that I didnt fly, this was a TP from my H1 sim to my H4 sim.

  74. Anny Helsinki says:

    bad word : [Resolved]

  75. U M says:

    The problem with the idea of bug searching is some people dig too deep and without meaning. I think we all learned from a year ago and all the false bug reports people sent in. Resulting in wild chases which lead the debuggers in countless waste of time and human resources. Atleaqst these days they have better sence and better staff. We are in alot better situation now when 1+ years ago.

  76. Sean Heying says:

    I am now able to cause the script stopping problem consistently.

    Attachments:
    Omicron v1.4,
    Amethyst Embedded Lock Slimline Collar,
    Huddle v1.1.3,
    Mystitool v1.0.21

    Second Life 1.19.0 (1) Feb 6 2008 15:13:43 (Second Life Release Candidate)

    You are at 200322.2, 236922.1, 411.9 in xxxx located at sim5585.agni.lindenlab.com (8.2.34.141:13006)
    Havok4 Beta Server 1.19.0.79349

    CPU: Dual i386 (Unknown) (2160 MHz)
    Memory: 2048 MB
    OS Version: Darwin 9.1.0 Darwin Kernel Version 9.1.0: Wed Oct 31 17:46:22 PDT 2007; root:xnu-1228.0.2~1/RELEASE_I386 i386
    Graphics Card Vendor: ATI Technologies Inc.
    Graphics Card: ATI Radeon X1600 OpenGL Engine
    OpenGL Version: 2.0 ATI-1.5.18
    LLMozLib Version: 1.1.0 (Mozilla GRE:1.8.0.12_0000000000)
    Packets Lost: 629/9843 (6.4%)
    Viewer Digest: 955b5d24-41ab-4e39-5c54-b43c204533dd

    My Av:

    My Home (H1) -> Armidi (H4) – Collar stopped working, Mystitool stopped working
    Armidi (H4) -> My Home (H1) – Collar stopped working
    Avpulli (H1) -> My Sandbox (H4) – Collar stopped working
    My Home (H1) -> My Sandbox (H4) – Collar stopped working

    My Alt:

    His Home (H4) -> My Home (H1) – Collar stopped working, Omicron fell off
    My Home (H1) -> His Home (H4) – Collar stopped working

    I have been able to fail just once with Omicron and Amethyst only… with the mystitool and huddle added the number is now 100% in 2 hours solid testing over two avatars.

    Earlier testing and comments about H1 to H1 seem to be my mistake, I was not checking every attachment after each TP. H1 to H1 cannot be duplicated.

    I’m not sure if it’s the number of prims I am carrying or the number of scripts. I feel it could be the sheer number of scripts I carry around with me 😉

  77. Storyof Oh says:

    problems with SL solved? am i alone in not being able to login on old or new release viewers?

  78. Cliff Dieffenbach says:

    Sidewinder…

    First, thanks for a response. I sent you as well a developer a notecard previously, but I suppose neither of you got it lol, which is ok. I’d rather him do the JIRA since he made it, but hopefully this will at least narrow down/fix another bug in the engine.

    SVC-1488 for Prim.Docker

    If you can’t get a hold of the developer (name in the JIRA) and/or need a demo let me know.

    Cliff

  79. Sidewinder Linden says:

    @72 Argent: If you’d like to talk about version string returns, it would be best to discuss this at one of the in-world Havok4 office hours so that we can talk about this. It has been discussed at some length, and I hesitate to rehash the issues here… The concept is definitely useful – how it gets implemented and when is an open question, as well as whether we’d all be left with legacy syntax that we’d rather not be left with if we rush something into place… /Sidewinder

  80. Sidewinder Linden says:

    @76 Sean: If you remove the collar, do you still have issues – it seems that’s a consistent thread (and I wonder if it might be a scripting problem in the collar) since I’ve done a pretty large number of teleports with MystiTool and Omicron without issues. Yes, the number of scripts does increase the load and possibility of issues… I rarely see someone with that many huds on constantly. Please contact me inworld to continue this discussion, so that we don’t use up large numbers of blog comment spots for this one potential bug, ok? Thanks, Sidewinder

  81. Sidewinder Linden says:

    @78 Cliff: Which version of Prim.Docker are you using? Was this test on the most recent deploy of Havok4 – in other words it was tested on or after last Friday? Thanks, Sidewinder

  82. Sidewinder Linden says:

    @69 Shadowquine: We have a new build of the simulator that is in QA at the moment. We believe that the locked no-copy rez problem you mentioned is now fixed, and will be deployed with that build. IF we pass QA on the first try, I hope that we’ll be able to deploy this new release by later in the week. When you see a blog post about the new build deploy please re-test to be sure it’s taken care of, ok? Thanks, Sidewinder

  83. Pingback: efecto antabuse

Comments are closed.