Jump to content

JoyofRLC Acker

Resident
  • Content Count

    160
  • Joined

  • Last visited

Posts posted by JoyofRLC Acker

  1. The 10 day thing is normally Tuesday I believe.  I am still not clear what maintenance was scheduled for today.

    I do have an explanation for what was happening in Blake this afternoon (EDT) ... we had requested that a couple of down sims in Blake be restarted (at about 11:30 SLT or thereabouts.  I gather that the CSR noticed a number of sims had not been restarted for 10 days and decided to do unscheduled restarts on all of them two, thus causing us to abandon our racing.

    However this has nothing to do with whatever was started at 9 am SLT.

     

  2. 6 hours ago, animats said:

    The "fix" that affected the main grid has to do with objects, not avatars, crossing. So it looks like the big winners there are people with large vehicles. I'm not seeing any improvement on the main grid for motorcycles and a small helicopter. Still getting random fails just driving around.

    The faster crossings on the beta grid AWS areas work fine.

    If you're a vehicle scripter, please take a look at BUG-229231 and see if this would affect you. Thanks.

    The improved performance seems to be persisting.   I'm measuring the "freeze time" at the border (as opposed to the agent detach / attach time), and that is regularly under 0.4 seconds now, usually under 0.2, versus 4 - 7 seconds before (on a Trudeau 12m yacht - 92 LI, 992KB script memory plus 400KB for HUDS).

    The other thing that seems to have changed is that when the agent fails to re-attach, I'm finding that I can TP back on to the boat with about 90% success rate (cam is a little messed up and you're stuck in mouselook mode, but the boat is responsive to gestures etc) - previously the success  rate of getting back on was < 20% for me.

    The bottom line is that we now have a very much higher % of boats finishing a race, and many more without falling off.

  3. Not sure if this is the place but ....

    Since last Tuesday / Wednesday there has been a dramatic improvement* in region border crossing performance in Agni (yes Agni, not Aditi).  Im not aware of any server code upgrade this week, just the "backend maintenance".

    So what's changed?

    And why have I not seen other comments on this?

    * The frozen time at the border is now almost zero in my Trudeau 12m yacht (not the lightest boat) whereas it 5 - 7 seconds before.   A quick test over a dozen sims with a time says its under 0.2 seconds; extensive racing over the last few days my subjective (i wouldn't race with a fine grained timer running) observation is almost all crossings are now sub second.

  4. So here's an odd thing.  While this "yay cloud go cloud" euphoria is building here, back in boring old Agni, Blake Sea has been performing since Tuesday afternoon ... wait for it ... amazingly well.  Yes, that's right  - AMAZINGLY WELL.

    Subjective observations of region border crossing freeze times in a Trudeau 12m yacht (NOT the slickest border crosser by a mile) are about a second (versus 5 - 7 seconds previously).  Chance of completing a race has gone way up to around 90%.   I havent fallen off in the last six races (3 on Shields, always reliable, and 3 on 12m).

    Im not at all sure what's changed - supposedly Tuesday there was no server update, just "backend maintenance" (whatever that is; and which by the way caused total chaos during Tuesday morning, the asset server seemed to be running on molasses, very long times to logout and even longer to log back).

    Can someone clarify what code versions are running on Aditi vs Agni?

  5. 4 hours ago, Aishagain said:

    @JoyofRLC Acker you say:

    I'm no apologist for LL but the work was clearly flagged up on the Grid Status page over the weekend, so I don't quite understand your reference to "hidden"?

    It was under past incidents.  I don't normally look there for info about whats going on currently.   I think that was technically the correct place, as they had posted that the maintenance was "finished".    The "" are due to the fact that Blake Sea was a mess yesterday; we had to abandon racing as I said.  My guess is that it was asset server not simulator issues.  Anyway ...

    Maybe they need 'recently completed and not yet fully checked out' section or something nearer the top.  More realistically, I guess I know I have to look there as well as the in progress maintenance.

  6. Blake Sea (Agni) is a mess.  And I think other sims too.

    It looks to me like no server updates today.   The usual Main Channel restarts for 'stale' sims.
    There was a hidden reference to "backend maintenance" this morning that supposedly ended at 10:30 SLT.

    There appear to be restarts going on now - very late - AGAIN!!!
    SL is the only major online system I know of that basically is scheduled to be fully operational for only 5 / 7 days (Tuesdays and Wednesdays are shot), minus any unscheduled down time.  It really is pretty pathetic.

    We had to abandon racing.   It looks to me like maybe asset server issues.   Objects super ultra slow to rezz (and then a bunch all rezz together in a hurry), when on board the boat scripts etc respond quickly, also comms between objects in the sim all operating well.  Problems at sim borders, crossing, or TP into sim.

    Another symptom is if you logout, it takes a long time for the logout to complete, and then after that a significant delay before you can log back in.

    Maybe someone can interpret all this?

  7. Speaking of 'telemetry' ... Ive always wondered how much resource goes into making those huge log files.

    ...

    Seems to me the priority right now is identifying actual bugs, and lets leave the performance evaluation until this goes to production.  (Unless they want to create a non-cloud Aditi ...)

  8. 6 hours ago, Lucia Nightfire said:

    Here is some good/interesting news, though.

    Mauve is back up on aditi and running in the cloud.

    I have a bunch of various speed/integrity tests I've made over the years.

    Here are two speed tests that involve llParseStringKeepNulls() vs llSubStringIndex():

    https://gyazo.com/c3c98a5e72a9e7c77978fc534de348ce

    I'll spare the technical details of what work these actually do repeatedly.

    On the left we have the tests running in Animesh Adult on agni in a full sandbox region with few running scripts, low script eps and low active objects in the region.

    On the right we have the tests running in Mauve on aditi in the cloud in a full sandbox region with many more active objects, more running scripts and more script eps.

    Ignore the viewer FPS on the left as I had both instances running at once and the one on the right was in focus.

    Notice that the performance of the same scripts in the cloud region is 2.5 - 3x faster despite it being a somewhat more busy region.

    Good work, and interesting.

    Is there a hypothesis as to the reason?  Has this been repeated yet, or is it just a one off?

  9. Leonardis - thats great news.  A more aggressive test than my sailing Simon's easter egg course in a Trudeau 12m.

    Im sorry if Ive missed it but do we know if there are any code differnces between these servers and Agni - and if so what.  If there are also code improvements as well as cloud servers Id be cautiously optimistic.  If the code is the same Im sceptical.

    However fwiw, Blake Sea has been very good last two days for our races (from Arabian line yesterday and Baltic today) - so yes before and after todays maintenance.

    Maybe Agni is trying not be shown up by Aditi!!  😄

  10. Quote

    c is actually backwards - we use a little more http than before (sorry to burst the anti-http bubbles)

    Not sure I understand what you are saying here.

    I will see what I can do about a fleet race or cruise (the latter would likely get larger numbers) may take a few days.

  11. Oops I should have added that the course was plain sailing in terms of server performance, all sim crossings were very smooth.  To the point I almost forgot the purpose of the exercise!!

    At the risk of sounding like a broken record I would like to know the relative contributions to that of:

    a) server code versions
    b) cloud servers
    c) http turned off
    d) ADITI is unpopulated and mostly unbuilt, very low script load etc.

    My money is on (d).

    Should we try a large fleet cruise?   And invite a bunch of tourists into Sirens Isle?  (kidding)

     

  12. Just completed the course:

    [13:12:30] Blake Sea Challenge: Progress: 100%
    [13:12:30] Blake Sea Challenge: Yay all done!!   It took 37 minutes and  you only got lost 7 times!
    Was NOT Lost - I was tacking (or gybing)!!

    For sail boats I would suggest wind direction NW NW SW SE that sort of thing.  I use N wind which led to some busy tacking up the west side of Windlass!

    Simon - great idea - thank you! 

     

  13. 1 hour ago, Simon Linden said:

    We just did some updates this morning (Monday July 27th) that fixes a bunch of the region connectivity issues.    I just did a tour and didn't hit any invisible walls.

    Along those lines, I made a quick "Blake Sea Challenge"   Go to secondlife://Aditi/secondlife/Morris/200/207/34  on the BETA aditi grid, and click on the red egg-shaped thing to try it out.   It will give you the "Blake Sea Challenge"  ... wear it and touch, and it'll get you going.   Follow the instructions to sail / fly / motor around 46 regions without doubling back and see how it goes.   Have fun and keep letting us know how it goes!

    Quote

     

    Suggested wind direction?

  14. 6 hours ago, Henri Beauchamp said:

    Greetings,

    I just discovered this new testing ground yesterday and gave it a try. While flying (avatar ”Fly”, i.e. not flying any scripted plane or such), most Blake* regions crossings were flawless (with at worst a split second ”hiccup” at crossing), but I did notice strange things, such as regions visible/active on map and invisible/inactive on mini-map and not accessible via crossing from other regions (while it was possible to TP into some). It was late in the night for me, however, so I did not take any note of time and region names.

    Today, I gave it another try and stumbled upon the same issue as yesterday, with one region in particular, inaccessible via borders crossing (excepted from one region: more on this later), but into which you can TP to. This region is ”Half Hitch” (debugging it might turn into a ”full itching” for you 😜). After I TPed into it, wearing a mildly scripted avatar (with 25 attachments containing around 80 well scripted/optimized scripts, consuming less than 3Mb or memory), I succeeded in crossing the South border to ”Turnbuckle”, but all other borders in the latter were uncrossable, so I headed back North to ”Half Hitch” and found myself ”stuck” just after the border, with the viewer disconnecting after timeout. When I re-logged to ”Last location”, I found myself back inside the region where I departed (”Kraken”) for my trip and before I TPed to ”Half Hitch”.

    Here is a zip file (too bad this forum refuses zip attachments...) with two logs (and a screen shot taken while ”stuck” and before timeout); the first log was from a session when I used a much lighter avatar with only a few attachments and scripts: I stopped in ”Kraken” to change for an ”heavier” outfit and since I got a ”Failed to rez attachment” message (not unusual in Aditi), I re-logged to avoid any possible issue with ”in flight” attachments data while crossing sims. The second log is the most interesting one, in particular with the warning message appearing here:

    
    2020-07-26T13:51:07Z INFO: process_crossed_region: Crossing region boundary
    2020-07-26T13:51:07Z WARNING: LLWorld::addRegion: Region exists, but old host 54.244.7.15:13006 does not match new host 0.0.0.0:0. Removing old region and creating a new one.
    2020-07-26T13:51:07Z INFO: LLWorld::removeRegion: Removing region 290048:268800
    2020-07-26T13:51:07Z INFO: LLViewerObjectList::killObjects: Removed 8282 objects for region Blake Sea - Half Hitch in 24.416ms

    It looks like a bogus IP (0.0.0.0) was transmitted to the viewer for the region it was entering !

    What you are describing is broadly similar to what I experienced in (or rather outside lol) Indefatigable the other day.

×
×
  • Create New...