Chat log from the meeting on 2013-04-09

From OpenSimulator

Revision as of 12:13, 9 April 2013 by Justincc (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
[11:00] Nebadon Izumi is online.
[11:00] Second Life: Region found!
[11:00] Second Life: Region found!
[11:00] Second Life: Region found!
[11:00] Second Life: Region found!
[11:00] Second Life: Teleport completed from hop://login.osgrid.org/jump4000/128/128/27
[11:01] Second Life: Region found!
[11:01] Andrew Hellershanks is online.
[11:01] Second Life: Teleport completed from hop://login.osgrid.org/jump8000/128/128/28
[11:01] Robert Adams: Greetings all
[11:01] Vivian Klees: Socrates sipped it first, he don't look too good atm
[11:01] BlueWall Slade: hehe
[11:01] Unknown User: Hi Robert
[11:01] moonfairie darkrose: hello everyone
[11:01] Justin Clark-Casey: hi robert, folks
[11:02] Andrew Hellershanks: Helllo everyone
[11:02] Unknown User: Hi Moon, Justin and Andrew
[11:02] BlueWall Slade: Hi Justin, Moonfairie
[11:02] BlueWall Slade: Andrew
[11:02] dave baxton: hello every one
[11:02] BlueWall Slade: hi dave
[11:02] Teravus Ousley: hello
[11:02] Unknown User: Hi Dave and Joe
[11:03] BlueWall Slade: Hey Joe - wher you goin with that gun in your hand?
[11:03] Unknown User: :_)...he is cloud top me so not seeing the gun
[11:03] Unknown User: top=to
[11:03] Unknown User: Hi Fred
[11:04] Unknown User waves
[11:04] logger sewell: Hi all
[11:05] Justin Clark-Casey: So just to ask, has anybody seen any problems where some objects will disappear without any obvious reason?
[11:05] Justin Clark-Casey: disappear from persistence, I mean
[11:05] Andrew Hellershanks: Yesterday was an interesting day. Just upgraded a grid from 073 to 075PF
[11:05] BlueWall Slade: from the database, justin?
[11:05] Justin Clark-Casey: bluewall: yes
[11:05] Nebadon Izumi: I havent seen it recently
[11:05] Unknown User: @justin no
[11:05] BlueWall Slade: wow
[11:05] Nebadon Izumi: but i have seen it
[11:06] Teravus Ousley: I've heard anecdotes about data loss like that.. but I haven't been able to recreate it.
[11:06] BlueWall Slade: I haven't, but that would be bad
[11:06] Justin Clark-Casey: nebadon: what kind of objects?
[11:06] Nebadon Izumi: generally I think it happens when people build to fast
[11:06] Nebadon Izumi: lots of people linking and unlinking simultaneously
[11:06] Nebadon Izumi: maybe even a mega linkset
[11:06] Unknown User: I modified a shoe the other day took it in inventory, rewore it and all was fine. On a relog all the pieces were gone except the sculpted foort
[11:06] Justin Clark-Casey: yes, that's possible
[11:06] Joe Radik: Gun in my hand?
[11:06] Nebadon Izumi: also
[11:06] Justin Clark-Casey: however, this is for objects which haven't changed for a long while
[11:06] BlueWall Slade: so, do you think they ever get persisted to start with?
[11:06] Nebadon Izumi: another thing I have been seeing lately
[11:06] moonfairie darkrose: does anyone know where i could get core''s Service bundle package at for groups and such
[11:06] Nebadon Izumi: if you shift drag a copy of something
[11:06] Nebadon Izumi: the copy will jump back on top of the original
[11:07] Nebadon Izumi: and it will look like it poofed completely
[11:07] Justin Clark-Casey: bluewall: yes they do - but at some point they disappear
[11:07] Nebadon Izumi: but actually its just stacked on other copy
[11:07] Justin Clark-Casey: bluewall: one possibility is that they somehow get flagged as temporary and so removed from persistence
[11:07] UUID Speaker: Pathfinder.Lester @pathlandia.dlinkddns.com: 29a47905-d904-4329-b2c9-8d53192c419e
[11:07] BlueWall Slade: any ideas about what is causing that?
[11:07] BlueWall Slade: ohh, ok
[11:08] Justin Clark-Casey: bluewall: unfortunately not - the temp thing is the only thing I can think of. Another possiblity is that the objects I'm concerend with are moving, so I'm wondering if they get flung to some bad location, flagged ast emporary and then deleted
[11:08] Justin Clark-Casey: but there's no evidence of that happening
[11:08] Justin Clark-Casey: nebadon: first I've heard of that, I have to say
[11:08] BlueWall Slade: the land doens't have a return time set does it?
[11:08] Nebadon Izumi: its happening to me a lot
[11:08] Nebadon Izumi: especially if I shift drag multiple objecsts at once
[11:09] Nebadon Izumi: i tend to build fast though
[11:09] Justin Clark-Casey: bluewall: I haven't checked that thoroughly but I'd be surprised if that was the case.
[11:09] BlueWall Slade: the movement - is is non-physical?
[11:09] BlueWall Slade: it it*
[11:09] Justin Clark-Casey: nebadon: sounds like a race condition
[11:09] Justin Clark-Casey: bluewall: yes
[11:09] BlueWall Slade: ok
[11:09] Nebadon Izumi: well actually I think it might be latency related
[11:09] Nebadon Izumi: this server I am working on is in Germany
[11:09] BlueWall Slade: I will setup something like that to check it on my grid
[11:09] Nebadon Izumi: i get like 120-180ms ping
[11:09] Nebadon Izumi: i have never noticed it working locally
[11:10] Nebadon Izumi: and it seems worst at certain times of the day
[11:10] Nebadon Izumi: like late afternoon early evening
[11:10] Nebadon Izumi: i notice it the most
[11:10] Justin Clark-Casey: bluewall: I'm thinking I might have to create a module to record all deletes
[11:10] Justin Clark-Casey: to at least know when this is happening
[11:11] Nebadon Izumi: Justin does the item poof after sim restart?
[11:11] Nebadon Izumi: or it is gone without any restart?
[11:11] BlueWall Slade: or mayb epyt some logging in the database layer?
[11:11] Nebadon Izumi: if its a restart, its likely never being added to the database
[11:11] Justin Clark-Casey: nebadon: I don't have enough evidence to say. Bu tthese servers are restarted daily
[11:11] Justin Clark-Casey: bluewall: that gets noisy quickly
[11:11] Nebadon Izumi: how are they being restarted?
[11:11] Justin Clark-Casey: nebadon: no, it's definitely added
[11:11] Justin Clark-Casey: but at some poitn it gets removed
[11:11] Nebadon Izumi: that might be part of your problem
[11:11] BlueWall Slade: well, that wouldn't really tell you what was asking for the deletion either.
[11:11] Nebadon Izumi: huh
[11:11] Nebadon Izumi: ya that is bizzare then
[11:11] Nebadon Izumi: thats not good
[11:12] Justin Clark-Casey: tell me about it :)
[11:12] BlueWall Slade: have you only seen it on the moving objects?
[11:12] Teravus Ousley: Yes, it will likely get noisy..
[11:12] Nebadon Izumi: i have never seen anything like that myself
[11:12] Justin Clark-Casey: that's what I was thinking, but apparantly another non-mover has disappeared as well. But that could be bad data for all I know
[11:12] Teravus Ousley: One suggestion that I've heard of is to write an object state every couple of minutes and when something goes missing, compare.
[11:13] Justin Clark-Casey: unfortunately, non-programmers never seem to appreciate the importance of narrowing things down and being accurate :)
[11:13] BlueWall Slade: I haven't missed anything like that
[11:13] BlueWall Slade: I will put soem moving objects out and watch them
[11:13] BlueWall Slade: ohh, lol
[11:13] Justin Clark-Casey: teravus: yeah, even checking number of prims on restarts would help here, presuming that active creation isn't goin on
[11:13] BlueWall Slade: (sometimes even programmers)
[11:13] Justin Clark-Casey: bluewall: thanks
[11:13] Nebadon Izumi: so wait these prims are moving?
[11:14] Justin Clark-Casey: yes, through llSetPos()
[11:14] Nebadon Izumi: its a NaN then
[11:14] Justin Clark-Casey: based on an initial positio nretrieved from llGetPos()
[11:14] Nebadon Izumi: shooting it off the sim
[11:14] Justin Clark-Casey: what makes you say that
[11:14] Nebadon Izumi: i have seen that
[11:14] Nebadon Izumi: with my train
[11:14] Teravus Ousley: oh noes, NaN blackholes
[11:14] Nebadon Izumi: i missed the its moving part
[11:14] Nebadon Izumi: sorry
[11:14] BlueWall Slade: if it was off the sim, then it would still be in the table, right?
[11:14] Justin Clark-Casey: nebadon: this is non-phys movement though
[11:14] Nebadon Izumi: my guess is its being fed a NaN
[11:14] BlueWall Slade: it would just have some crazy pos
[11:14] Nebadon Izumi: and being returned
[11:14] Pathfinder's Hypergrid Adventurer Hat: Current Hypergrid Coordinates: 10000,10000
[11:15] Justin Clark-Casey: I didn't think that would return an obj
[11:15] Nebadon Izumi: hmm I have seen it happen with my train
[11:15] BlueWall Slade: I think it would only if the object were scripted to die when it moves off world
[11:15] Justin Clark-Casey: nebadon: was there any associated physics log message?
[11:16] Nebadon Izumi: off the top of my head I can not recall
[11:16] Justin Clark-Casey: I think there are circumstances where such an object would be marked as temporary
[11:16] Justin Clark-Casey: I looked through the associated code and there's no obvious NaN problem
[11:16] Nebadon Izumi: its been a while since i played with the train script
[11:16] Justin Clark-Casey: but I didn't do that forensically
[11:16] BlueWall Slade: Ever see "The Never Ending Story"?
[11:16] BlueWall Slade: hehe
[11:16] Nebadon Izumi: you know
[11:17] Nebadon Izumi: justin if your region has a neighbor
[11:17] Nebadon Izumi: and it goes over the border
[11:17] Justin Clark-Casey: no neighbours
[11:17] Nebadon Izumi: it might get removed from DB
[11:17] Nebadon Izumi: ok hmm
[11:17] Justin Clark-Casey: yeah, I preobably have to get a module to log removal to get a handle on this
[11:17] Nebadon Izumi: ya it should be fairly obvious if you do that I would think
[11:18] Nebadon Izumi: moving objects can be very erratic
[11:18] Justin Clark-Casey: it would help, rather than guessing from too little data
[11:18] Nebadon Izumi: my train does all kinds of wierd things
[11:18] BlueWall Slade: I would look to see what the object perms on the land are set to too
[11:18] Nebadon Izumi: sometimes if flips upside down for no reason
[11:18] Nebadon Izumi: it wil go sideways, all kinds of odd things
[11:18] Justin Clark-Casey: and that's not physical movement?
[11:18] Nebadon Izumi: correct
[11:19] BlueWall Slade: it's looking for the track?
[11:19] Justin Clark-Casey: I didn't fix a bunch of minor raec conditions associated with vectors recently but it wouldn't surprise me if there are more
[11:19] Teravus Ousley: space conditions...
[11:20] Nebadon Izumi: http://forums.osgrid.org/viewtopic.php?p=16845#p16845
[11:20] Nebadon Izumi: here is the script
[11:20] moonfairie darkrose: ok have a question about Core's Service Bundle is there any place to get it at all or is it gone totally
[11:20] Nebadon Izumi: are avatars sitting on this object?
[11:20] Justin Clark-Casey: nebadon: no
[11:20] Nebadon Izumi: the one your having trouble with?
[11:20] Nebadon Izumi: ok
[11:21] Justin Clark-Casey: moonfairie: somebody on the ml said that core had left the scene
[11:21] Justin Clark-Casey: I mean, OpenSimulator, etc., so I don't know where one would get the bundle now
[11:21] dave baxton: is thare a replacement for it anyone no
[11:21] Unknown User: saw it yet about 3 weeks ago Moon
[11:21] Teravus Ousley searches for drama
[11:21] BlueWall Slade: those are not that hard to install anyway
[11:21] Justin Clark-Casey: nebadon: it's just an object which bobs up and down at semi-random intervals
[11:22] Justin Clark-Casey: nebadon: very simple really
[11:22] Nebadon Izumi: ok ya I might even know the objects your talking about
[11:22] BlueWall Slade: that is all the movement , just bobbing?
[11:22] Justin Clark-Casey: yes
[11:22] BlueWall Slade: ok, that's easy
[11:22] BlueWall Slade: I'll put some out
[11:22] Justin Clark-Casey: thanks
[11:23] BlueWall Slade: neb, that train script uses waypoints, so it's not searching for the guide
[11:23] Nebadon Izumi: correct bluewall
[11:23] Nebadon Izumi: which is more likely how justins problem item is too
[11:23] BlueWall Slade: the VRC script uses llSetPrimitiveParamsFast
[11:23] BlueWall Slade: so, it's non phys too
[11:24] Nebadon Izumi: I experimented with that on this train
[11:24] BlueWall Slade: but, it doesn't run around un-manned
[11:24] Nebadon Izumi: and found it had same problems
[11:24] Nebadon Izumi: it was very random
[11:24] Nebadon Izumi: it would run for days fine
[11:24] Unknown User: justin: is there a mantis and object description/script
[11:24] Nebadon Izumi: then all the sudden out of the blue its upside down
[11:24] Nebadon Izumi: or just gone from the sim completely
[11:24] BlueWall Slade: that is strange
[11:25] Nebadon Izumi: ya or totally sideways
[11:25] Teravus Ousley doesn't find any
[11:25] Nebadon Izumi: very strange for sure
[11:25] BlueWall Slade: I would expect that it is getting some bad data for a waypoint to make that happen
[11:25] Nebadon Izumi: i know its not the script itself
[11:25] Justin Clark-Casey: there's no mantis yet. I wanted to do more investigation first
[11:25] Nebadon Izumi: that script ran in SL for probably 4-5 months straight
[11:25] Nebadon Izumi: without the sim ever restarting
[11:25] BlueWall Slade: that must have been a long time ago, lol
[11:26] Nebadon Izumi: ya that was before OpenSim existed
[11:26] Nebadon Izumi: hehe
[11:26] Justin Clark-Casey: anyway, perhaps we should ask if there are any other openSimulator topics today?
[11:27] BlueWall Slade: I looked at moving teh avatars with llSetLinkPrimitiveParams, but didn't do any actual work yet
[11:27] Kayaker Magic: I have a question, in SL they say scripts are the lowest priority thing that the simulator does
[11:27] Kayaker Magic: is that true in OpenSim as well?
[11:27] Nebadon Izumi: probably not no
[11:27] Justin Clark-Casey: kayaker: no, I think because the architecture is fundamentally different
[11:27] moonfairie darkrose: is there something similiar like core's bundle available? we are trying to get our grid up and running with hypergrid enabled
[11:28] Justin Clark-Casey: I think in SL a lot more stuff is run on a single thread effectively, such as physics and script udpates
[11:28] Nebadon Izumi: the simulator in SL is very very different
[11:28] Kayaker Magic: So a bad script can lag a sim here, while in SL they say it cannot?
[11:28] Nebadon Izumi: its likely not 1 monolithic process
[11:28] Justin Clark-Casey: in OpenSimulator they are completely independent
[11:28] Justin Clark-Casey: kayaker: I think that's possible
[11:28] BlueWall Slade: yeah, I think they queue the scripts and insert them into the engine
[11:28] BlueWall Slade: like threads
[11:29] Justin Clark-Casey: moonfairie: I don't think so at this point
[11:29] Justin Clark-Casey: moonfairie: one would have to add the components individually, like groups
[11:29] BlueWall Slade: you can get the profiles, search and groups from their repositories
[11:29] Nebadon Izumi: groups is in core[
[11:29] Nebadon Izumi: on master git
[11:29] Unknown User: doesnt Diva's wifi have it?
[11:30] Justin Clark-Casey: I believe so
[11:30] BlueWall Slade: I haven't tried those yet
[11:30] Nebadon Izumi: yes but divas wifi is hobbled
[11:30] Nebadon Izumi: to only 2 groups
[11:30] Nebadon Izumi: until next release
[11:30] moonfairie darkrose: yes ive read that too
[11:30] Key Gruin: Rich
[11:30] Justin Clark-Casey: I don't believe the code in OpenSimulator has been much used yet, unless anyone can say different
[11:30] Justin Clark-Casey: the new groups code, that is
[11:30] Andrew Hellershanks: Does groups in master eliminate need for any group stuff on webserver now?
[11:30] Richardus Raymaker: hi key
[11:30] Nebadon Izumi: probably not large scale no
[11:31] Justin Clark-Casey: andrew: it would
[11:31] Nebadon Izumi: but a few people have tested it
[11:31] Andrew Hellershanks: justin, nice. I should run some tests on that
[11:31] Justin Clark-Casey: but group stuff on a webserver might still perform better. Hard to say
[11:31] Unknown User: there is another one out there that i forget the name of
[11:31] Justin Clark-Casey: only becasue of the greater efficiency of Apache vis a vis the C# webserver, perhaps
[11:31] Unknown User: partly free
[11:31] Nebadon Izumi: you could always break groups into its own robust console too
[11:31] Andrew Hellershanks: web based groups is ok until you get to around 100 or so in the group
[11:31] Nebadon Izumi: that would give it a bit more power
[11:31] Justin Clark-Casey: apache + php, that is
[11:31] Nebadon Izumi: probably even run several groups robust consoles and load balance with nginx
[11:32] Nebadon Izumi: i am speculating but i see no reason why it could not be done that way
[11:32] Justin Clark-Casey: I see no reason it couldn't either
[11:32] moonfairie darkrose: we are trying to start our own grid but are not very familiar with all that
[11:32] Justin Clark-Casey: doing a grid at any scale is a very complex task at this time
[11:32] Nebadon Izumi: ya there is a lot to learn to running a full grid
[11:32] BlueWall Slade: if it gets down to that, then it might be better to keep it in apache?
[11:32] Tiffany Magic: Kayaker... please bring up the problem you have been talking to me about?
[11:33] Richardus Raymaker: Andrew. sofar i know groups in master is limited to ?? groups ?
[11:33] Richardus Raymaker: ALso not sure if i offline IM the sorting is fixt
[11:33] Justin Clark-Casey: there's no limit to group numbers in master
[11:33] Richardus Raymaker: Also i found a possible small problem with rotation but tahts someting for later now.
[11:34] Justin Clark-Casey: I don't know how well the offline IM stuff works =- have no tried it
[11:34] Nebadon Izumi: group limits are viewer side I think
[11:34] Andrew Hellershanks: Richardus, I haven't looked at it yet. There is an OS setting about number of groups a person can join.
[11:34] BlueWall Slade: how about inventory, notices, permissions, etc.?
[11:34] Richardus Raymaker: i still usee php code. only heared things
[11:34] Nebadon Izumi: there is a server side and viewer side to that limit i think
[11:34] BlueWall Slade: V3 uses some code to search for groups
[11:35] Nebadon Izumi: hello Cuteulala
[11:35] Andrew Hellershanks: The limit I know is really just a value that can be sent as part of the login response packet to inform the viewer what the limit is on groups that can be joined.
[11:35] Cuteulala Artis: my ipod screwed :(
[11:35] Cuteulala Artis: wont take my password now disabled
[11:35] Justin Clark-Casey: you around, dan?
[11:35] Richardus Raymaker: ok. i try it. i have seen soem weird problem with the edit rotation
[11:36] dan banner: yes justin
[11:36] Richardus Raymaker: The y axe is limted between 270 -> 0 -> 90 degree.
[11:36] Justin Clark-Casey: dan: I saw your comment on the mantis but as far as I could see, lbsa was still set to SmartThreadPool
[11:36] Andrew Hellershanks: Oh, search... Any pointer as to where I would look in the OS code base to find where it does searches given part of an avatars name?
[11:36] Richardus Raymaker: i cant rotate it more then that with the arrows in the editor. imprudence have the same problem
[11:36] dan banner: ya i must have cancelled the edit instead of save
[11:36] Andrew Hellershanks: The search doesn't do partial matches.
[11:36] dan banner: question is QueueUserWorkItem or UnsafeQueueUserWorkItem?
[11:37] Justin Clark-Casey: dan: if you could switch that would be valuable. I really want to resolve this problem but it might have to wait till after my holiday now
[11:37] dan banner: wright is using QueueUserWorkItem
[11:37] Justin Clark-Casey: dan: I would say go for Queue seeing as this region seems pretty stable with it
[11:37] Andrew Hellershanks: Another holiday so soon Justin? :-)
[11:37] Justin Clark-Casey: bah, so soon? :)
[11:37] dan banner: okay, ya lbsa is set to QueueUserWorkItem now
[11:37] Justin Clark-Casey: dan: thanks
[11:38] dan banner: np
[11:38] Cuteulala Artis: :o
[11:38] Justin Clark-Casey: dan: those pastebins are still showing exactly the same thing - really puzzling
[11:38] Andrew Hellershanks: Doesn't seem like it was that long ago but time passes very quickly some times.
[11:38] dan banner: sorry for the confusion
[11:38] Cuteulala Artis: when are we gonna get the new lbsa
[11:38] Justin Clark-Casey: andrew: my last holiday was xmas :) I'll be away for about 1.5 weeks starting from tomorrow
[11:38] Justin Clark-Casey: so may or may not be at next weeks meeting
[11:38] Nebadon Izumi: ok
[11:38] Teravus Ousley: cool
[11:39] Nebadon Izumi: you traveling Justin?
[11:39] BlueWall Slade: justin - that texture is as pic of a hung avatar...
[11:39] Robert Adams: despite its scary name "UnsafeQueueUserWorkItem" is nearly equivalent to the 'safe' one with differences in termination/aborting
[11:39] Richardus Raymaker: neb , what viewer are you useing ? can you maby test a small thing ?
[11:39] BlueWall Slade: that budy part was detached from it
[11:39] Justin Clark-Casey: nebadon: I need to move house, which is rather bad
[11:39] Justin Clark-Casey: bluewalL; disturbingly, it's refusing to load for me...
[11:39] Teravus Ousley: a vaca every 5 to 6 months isn't bad!
[11:39] Andrew Hellershanks starts to think Justin gets a lot of time off in a year.
[11:39] Nebadon Izumi: I think it was dubbed unsafe because it doesnt work on older mono versions
[11:39] Justin Clark-Casey: oh there we go
[11:39] Andrew Hellershanks: I was lucky to have 3 weeks where I used to work.
[11:39] Nebadon Izumi: after mono 2.10 unsafe did seem to work ok on mono
[11:39] Richardus Raymaker: Robert, i read on MS page that UnsafeQueueUserWorkItem is not best thing to use. thats why i still avoid it
[11:39] Justin Clark-Casey: Nebadon: No, I think unsafe is actually something in the c# sdk
[11:40] Nebadon Izumi: ah well I do know on older monos it did not work
[11:40] Nebadon Izumi: like 2.6 and below I think
[11:40] Teravus Ousley: oh, that sucks.. house moving.. That's hardly a vacation.
[11:40] Justin Clark-Casey: ahh, its' called unsafe because it could allow for privliege esvcalation
[11:40] Unknown User: virtual house?
[11:40] Justin Clark-Casey: but in our case that's not an issue
[11:40] Justin Clark-Casey: teravus: I'm hoping it will only take 0.5 weeks (I'm a dreamer :)
[11:41] Justin Clark-Casey: so Unsafe is fine
[11:41] Teravus Ousley: Yes, "you're not the only one... "
[11:41] Nebadon Izumi: ya i remember the unsafe being much faster on .net though
[11:41] Nebadon Izumi: than smartthreadpool
[11:41] Justin Clark-Casey: ken was saying that unsafe was a lot quicker on windows
[11:41] Justin Clark-Casey nods
[11:41] Nebadon Izumi: it was considerbly faster rezzing the 45k prim klein bottle
[11:41] Teravus Ousley: I can hope that you manage it in 0.5 weeks also..
[11:41] Nebadon Izumi: like 3-4 times faster
[11:42] Cuteulala Artis: Need fix teraforming to slow :P
[11:42] Dahlia Trimble is online.
[11:42] Justin Clark-Casey: yes, I always wanted to look at why terraforming is slow. Still haven't had the time...
[11:42] BlueWall Slade: yeah we need to selectivly drop requests in that with some throttle
[11:43] BlueWall Slade: we should throttle the inputs
[11:43] Cuteulala Artis: Rev made teraforming real fast in aurora but i dont like aurora
[11:43] Andrew Hellershanks: terraforming seems slow before it starts to respond. Once it does, it seems ok
[11:43] Robert Adams: terraforming gets throttled by the output queue bandwidth
[11:43] Andrew Hellershanks: At first nothing seems to happen.
[11:43] Robert Adams: terrain updates don't get much bandwidth
[11:43] Nebadon Izumi: ya its even worst on mega regions
[11:43] Unknown User: i find some viewers are better then others for terraformin
[11:43] Cuteulala Artis: In aurora its instant responce all the time
[11:43] Nebadon Izumi: its almost impossible to teraform on a mega
[11:43] BlueWall Slade: Robery, we throttle everything at the output, but I think we accept inputs
[11:44] Unknown User: unhjcombine first
[11:44] Justin Clark-Casey: perhaps he made terrain updates a highe rpriority
[11:44] BlueWall Slade: so, we end up with a backlog in the queue
[11:44] BlueWall Slade: am I right??
[11:44] Teravus Ousley: MegaRegions, unfortunately where terrain is concerned, are a casualty of the throtting mechanism
[11:44] Cuteulala Artis: i cant wait till osgrid fixes that
[11:44] Justin Clark-Casey: terraforming isn't done very often
[11:44] Cuteulala Artis: will help me deeply
[11:44] Dahlia Trimble: hi
[11:44] Justin Clark-Casey: hi dahlia
[11:44] BlueWall Slade: Hi Dahlia
[11:44] Nebadon Izumi: really? terraforming isnt done a lot?
[11:44] Cuteulala Artis: i teraform every day
[11:44] Nebadon Izumi: the first thing any new person would likely do is terraform
[11:44] Justin Clark-Casey: I mean, as a proportion of all time spent in a simulator
[11:44] Cuteulala Artis: :p
[11:44] Teravus Ousley: .. main reason for megaregion terraforming problems is the throttle is restricted by an extra certain amount
[11:44] BlueWall Slade: Cuteulala, try to select the land and do one thing at a time with it
[11:45] Justin Clark-Casey: yeah, but they're not doing it every day, right?
[11:45] Teravus Ousley: .. in each child region
[11:45] Nebadon Izumi: yes but its one of the 1st things people do
[11:45] Justin Clark-Casey: unless you're cuteulala I expect :)
[11:45] Tiffany Magic: Austin... terrforming is done on every regions we put out. And changed and tweaked regularly.
[11:45] Nebadon Izumi: the only reason i dont do it alot
[11:45] Justin Clark-Casey: I'm not saying it's unimportant.. I would actually use that as an argument for increasing the output packet priority if that would help
[11:45] Nebadon Izumi: is because the in world tools suck so bad
[11:45] Nebadon Izumi: i use blender
[11:45] Tiffany Magic: *Justin
[11:45] Justin Clark-Casey: i.e. because it's not done constantly it won't affect ppl a lot
[11:45] BlueWall Slade: I remember the map was doing the same thing...
[11:45] Cuteulala Artis: i can teraform a sim in aurora in 5 mins flat a nice teraform in osgrid it takes a couple of hours
[11:45] Richardus Raymaker: aha mega region terraforming yes thats going slow if you go outside the sim you are with avatar or maby root :(
[11:46] Unknown User: terraforming is fastest on a locally hosted standalone
[11:46] BlueWall Slade: we throttled the network packets, but kept accepting input
[11:46] Teravus Ousley: so unless you're terraforming the root region, it's going to be extra slow terraforming the extended regions
[11:46] Nebadon Izumi: ya
[11:46] BlueWall Slade: so, if you zoom out, it would eat the sim alive
[11:46] Unknown User: save it and transfer over
[11:46] Cuteulala Artis: i think we limit osgrid then its capable i uploaded a 300 meg super mesh file a entire city from skyrim without errors :p
[11:46] BlueWall Slade: afk fro a few...
[11:46] Justin Clark-Casey: And the asset database quivered...
[11:46] Tiffany Sicling: I remember the days where terraforming was realtime and was finished when you lifted your finger off the mouse button
[11:46] Justin Clark-Casey: or file system rather for osgrid :)
[11:47] Cuteulala Artis: it onlt took 10 mins
[11:47] Nebadon Izumi: the viewer does most of the hard work for mesh upload
[11:47] Nebadon Izumi: linksets are worse
[11:47] Nebadon Izumi: really big linksets are the worst
[11:47] Justin Clark-Casey: cute: nice
[11:48] Cuteulala Artis: osgrid should have an option for run at your own risk lol
[11:48] Justin Clark-Casey: I will probably think about doing an OpenSimulator release after my hols
[11:48] Cuteulala Artis: id love to push my cores a bit harder they are being lazy at 0% most of time lol
[11:48] Kayaker Magic: I have a problem in Tiffany's grid where every time I save a big script, it takes longer to save. Eventually the sim statistics get bad, and a sim restart speeds things back up.
[11:49] Kayaker Magic: I'm hypothesizing that some system resource is getting lost because of the slow response of the distant servers. Any ideas?
[11:49] Justin Clark-Casey: but would really like to fix the ghost avatar issue first
[11:49] Justin Clark-Casey: kayaker: how big a script?
[11:49] Kayaker Magic: 800 lines
[11:49] Richardus Raymaker: oh.. the ghosts are back
[11:49] Robert Adams: should we just remove the terrain output bandwidth limit?
[11:49] Simulator Version v0.5 shouts: OpenSim 0.7.6 Dev          94d4414: 2013-04-03 00:48:36 +0100 (Unix/Mono)
[11:50] Justin Clark-Casey: kayaker: what version of opensim is that?
[11:50] Cuteulala Artis: with all do respect i think we need to fix stuff that eddects the users productivity before a ghost issue we need to fix teraforming and sitting that way osgird will feel more reliable to new users
[11:50] Justin Clark-Casey: robert: is that in the standard output packet throttles?
[11:50] Kayaker Magic: It was 0.7.3 until yesterday
[11:50] Richardus Raymaker: Robert. make the terrainbandwidth customizable
[11:50] Justin Clark-Casey: cute: well, the ghost issue is a regression, the other things are existing issues
[11:50] Cuteulala Artis: the most commong issues should be fixed first
[11:50] Andrew Hellershanks: Kayaker, that is happening in the grid since this past monday evening?
[11:50] Elisabeth Thorne: May I ask a stupid question?
[11:50] Justin Clark-Casey: they will get fixed but this takes time
[11:50] Cuteulala Artis: understood no rushy
[11:50] Justin Clark-Casey: elisabeth: sure
[11:50] Cuteulala Artis: lol
[11:50] Robert Adams: I thinks so, it's possible it was restricted to "what it should be" when the throttling was re-engineered a year ago or so
[11:51] Kayaker Magic whispers: Haven'had time to try crashing a sim since the upbrade!
[11:51] Justin Clark-Casey: robert: I think it would be fine to unlimit those or vastly increas ethe limit
[11:51] Robert Adams: a fancy fix would be to unlimit it when editting
[11:51] Unknown User: is ghosting issue being reported by other grids?
[11:51] Elisabeth Thorne: Why .NET/Mono? Has any thought been given to moving to, say, C++?
[11:51] Justin Clark-Casey: kayaker: that's very odd behaviour for a script upload
[11:51] Andrew Hellershanks: Robert, would be interesting to see if improves response time.
[11:51] nurbsPlane9: DynamicTextureModule: Error preparing image using URL http://images.intellicast.com/WxImages/Precipitation/usa.gif
[11:51] nurbsPlane9: DynamicTextureModule: Error preparing image using URL http://images.intellicast.com/WxImages/Radar/usa.gif
[11:51] nurbsPlane9: DynamicTextureModule: Error preparing image using URL http://images.intellicast.com/WxImages/CustomGraphic/wgfhi.gif
[11:51] nurbsPlane9: DynamicTextureModule: Error preparing image using URL http://images.intellicast.com/WxImages/Satellite/hiusa.gif
[11:51] Justin Clark-Casey: kayaker: it should take practically no resources
[11:52] Justin Clark-Casey: Elisabeth: unfortunately, OpenSimulator is 500,000 lines of source code
[11:52] dan banner: arielle: how many other grids run 0.7.6-dev?
[11:52] Justin Clark-Casey: which also rests on some more code in libomv
[11:52] Dahlia Trimble: c++? Id much rather move to a dynamic language
[11:52] Unknown User: ghosting was happening on .7.5 already
[11:52] Kayaker Magic: There is always a scary delayy when saving a script, the viewer freezes for a second.
[11:52] Justin Clark-Casey: writing in some other langauge would still take a vast amoutn of work
[11:52] Primitive: START EFFECTS
[11:52] Elisabeth Thorne: :) I'd rather anything than being stuck with Mono.
[11:52] Kayaker Magic: On that grid, the frozen viewer time gets longer and longer.
[11:52] Primitive: STOP EFFECTS
[11:52] Primitive: START EFFECTS
[11:52] Primitive: STOP EFFECTS
[11:53] Nebadon Izumi: is that you Teravus?
[11:53] Nebadon Izumi: the effects messages?
[11:53] Elisabeth Thorne: And like I said, it was a stupid question :)
[11:53] Andrew Hellershanks: Elisabeth, I would hate to see OS in C++. The few programs I have compiled that are written in C++ take ages to compile.
[11:53] Justin Clark-Casey: kayaker: that's very odd. You could try running on inbound http logging on the simulator and see if the prioblem is there
[11:53] Teravus Ousley: The effects messages are from the notecard suggestion box
[11:53] Nebadon Izumi: oh ok
[11:53] Justin Clark-Casey: Elisabeth: no worries. I would actually love to see rewrites in toehr languages but it is a lot of work
[11:53] Andrew Hellershanks: Python! ;-)
[11:54] Justin Clark-Casey: and this whole system is byzantinely complicated
[11:54] Tiffany Sicling: ASM would make it very fast ;)
[11:54] Elisabeth Thorne: Oh, I have no doubt of that. A shame there aren't better translation tools.
[11:54] Justin Clark-Casey: humans can practically never beat a compiler at generating asm
[11:54] Nebadon Izumi: the problem is sandboxing it
[11:54] Nebadon Izumi: heh
[11:55] Nebadon Izumi: you certainly would not want to give just anyone that level of access to script engine
[11:55] Dahlia Trimble: c++ runs faster but c# is written much faster
[11:55] Justin Clark-Casey: I shudder to think of the security issues if OpenSimulator were written in c++
[11:55] Andrew Hellershanks: The thought of doing something as big as OS in pure assembler is a very scary thought.
[11:55] Nebadon Izumi: ya
[11:55] Nebadon Izumi: you could literally allow someone to erase your entire hard drive
[11:55] Andrew Hellershanks: Dahlia, C++ may run faster but it compiles a whole lot slower.
[11:55] Nebadon Izumi: if you didnt protect against stuff like that
[11:56] Tiffany Sicling: c# has to have .NET/mono to even think about running
[11:56] Dahlia Trimble: Andrew, if compiles quite fast for me
[11:56] Elisabeth Thorne: Only the first time. Then you just recompile what has changed.
[11:56] Tiffany Magic: Justin: So do you have any suggestions for us to deal with the problem Kayaker was talking about?
[11:56] Dahlia Trimble codes a fair but of c++
[11:56] Dahlia Trimble: *bit
[11:56] Andrew Hellershanks: Sorry to hear that. ;-)
[11:56] Teravus Ousley: for large systems... c++ can be fast of the project is designed for it :)
[11:57] Justin Clark-Casey: tiffany: so what server stats are degrading over uplaods?
[11:57] Kayaker Magic: In my spare time, I'll do the test again in 0.7.5, get some better statistics.
[11:57] Andrew Hellershanks: Don't know why but the SDR code I was compiling yesterday too ages to build. So does Blender if I do full build. OS compiles rather quickly by comparison.
[11:57] Justin Clark-Casey: this would be an extremely odd problem because file uploads don't touch the scene loop at all, and take very few resources
[11:57] Kayaker Magic: Eventually SIM FPS and Physics FPS drop.
[11:57] Justin Clark-Casey: kayaker: after how many script changes?
[11:57] Kayaker Magic: If I don't reset the sim, it crashes.
[11:57] Justin Clark-Casey: crashes with what error?
[11:57] Nebadon Izumi: what is the problem exactly with scripts?
[11:58] Nebadon Izumi: could this be an app domain issue?
[11:58] Kayaker Magic: Just normal debugging, save, test edti save edit save edti....
[11:58] Kayaker Magic: loose track of time....
[11:58] Andrew Hellershanks: Nebadon, Kayaker is saying that the more you save scripts, the slower the region becomes.
[11:58] Nebadon Izumi: linux or windows?
[11:58] logger sewell: windows
[11:58] Nebadon Izumi: i bet its app domain setting
[11:58] Nebadon Izumi: oh hrmm
[11:58] Nebadon Izumi: i was hoping you said linux
[11:58] Richardus Raymaker: hmm. i never have seen that scipt habbit.
[11:58] Nebadon Izumi: it still could be app domain setting
[11:59] Kayaker Magic: Hmm, it does not happen on my linux sim
[11:59] Unknown User: Dan: I have seen 7.6 specific regions on other , mostly they are on Standalonesgrids
[11:59] Nebadon Izumi: the thing is if you have app domain = false
[11:59] Nebadon Izumi: scripts never unload
[11:59] Cuteulala Artis: I push my sims to there limits past the limits yet to crash one haha i just pure lucky i sapose
[11:59] Nebadon Izumi: so if you edit a script 500 times
[11:59] Nebadon Izumi: it will be loaded in memory 500 times
[11:59] Nebadon Izumi: Kayaker check a setting for me
[11:59] Justin Clark-Casey: I wonder how much more memory would be used by a big script
[11:59] Andrew Hellershanks: What should that seeting be for Windows boxes? I lose track of what is recommended.
[11:59] Nebadon Izumi: in [Xengine]
[11:59] Nebadon Izumi: check appdomain =
[11:59] Nebadon Izumi: make sure its set for true in windows
[11:59] Nebadon Izumi: and see if you still have that same problem
[12:00] Andrew Hellershanks: Nebadon, it is true
[12:00] Nebadon Izumi: the osgrid default is false
[12:00] Nebadon Izumi: ok
[12:00] Kayaker Magic: you mean set it true in my linix sim and see if it happens there?
[12:00] Kayaker Magic: I can do that.
[12:00] Nebadon Izumi: no
[12:00] Nebadon Izumi: well you can try in linux
[12:01] Nebadon Izumi: to true, but if you have over 500 scripts
[12:01] Nebadon Izumi: dont even bother
[12:01] Nebadon Izumi: unfortunately the best option in linux is false
[12:01] Nebadon Izumi: and in windows true runs much better
[12:01] Justin Clark-Casey: loading each script into a spearate AppDomain takes a long time on linux
[12:01] Justin Clark-Casey: but much less time on Windows
[12:01] Nebadon Izumi: and a lot more memory
[12:01] logger sewell: could it be that it is caching the scripts on windows
[12:01] Nebadon Izumi: infact this region wont even start with it set to true
[12:02] Nebadon Izumi: it gets about 50% loaded and stack traces
[12:02] Andrew Hellershanks: That seems odd. Why should it take longer to load in Linux? A mono issue?
[12:02] Justin Clark-Casey: logger: not really, since scripts are very small
[12:02] Justin Clark-Casey: andrew: yes, because mono (and linux) work diferently
[12:02] logger sewell: ok
[12:02] Andrew Hellershanks nods
[12:02] Nebadon Izumi: you can also try typing "force gc"
[12:02] Nebadon Izumi: on your sim console
[12:02] Nebadon Izumi: this tends to free up memory fast if you need it
[12:02] Justin Clark-Casey: which is one reason I haven't flipped that flag to false by default, much as I would like to
[12:03] logger sewell: I might have to run Kayakers region in a console by it self to record the data
[12:03] Nebadon Izumi: wait
[12:03] Nebadon Izumi: you have multiple regions running?
[12:03] Nebadon Izumi: in 1 simulator?
[12:03] Andrew Hellershanks: no
[12:03] Nebadon Izumi: that could also be the problem
[12:03] Nebadon Izumi: ok good
[12:03] Andrew Hellershanks: 1 region, 1 instance
[12:03] Nebadon Izumi: good
[12:04] Nebadon Izumi: thats the best way to do it
[12:04] Elisabeth Thorne: So AppDomain should be false for linux?
[12:04] Nebadon Izumi: correct
[12:04] Justin Clark-Casey: elisabeth: it's vastly quicker but will lead to a memory leak over time as new scripts are compiled
[12:04] Nebadon Izumi: yes, and like i said
[12:04] Nebadon Izumi: if you have 2000 scripts
[12:04] Nebadon Izumi: there is a good chance the region wont even start at all
[12:04] Nebadon Izumi: if its set for true in linux
[12:05] Justin Clark-Casey: They seem to start fine for me
[12:05] Nebadon Izumi: Wright Plaza wont start with it true
[12:05] Justin Clark-Casey: but do take a stupidly long time
[12:05] Nebadon Izumi: it stack traces before logins are open
[12:05] Justin Clark-Casey: have you tried that with mono 3 btw?
[12:05] Nebadon Izumi: no I actually have not
[12:05] Elisabeth Thorne: Startup speed doesn't concern me, it's stability.
[12:05] Justin Clark-Casey: not important, just wondering if anything has changed
[12:05] Nebadon Izumi: I should
[12:05] Nebadon Izumi: I will soon
[12:05] Fearghus McMahon: Hi everybody
[12:06] Nebadon Izumi: I would be surprised if that changed
[12:06] Nebadon Izumi: its possible though
[12:06] Nebadon Izumi: GC has improved
[12:06] Justin Clark-Casey: hello fearghus
[12:06] Richardus Raymaker: oh about mono 3, it seems sometimes you need to make symbolkic link from lib64 to /lib otherwise opensim never start
[12:06] Justin Clark-Casey: I would be surprised if it's changed too, tbh, but it's possible
[12:06] Nebadon Izumi: thats more likely a issue with your distro
[12:06] Dahlia Trimble: GC is still horrible
[12:06] Nebadon Izumi: than mono itself Richardus
[12:06] Richardus Raymaker: Not really, compiled mono in userspace
[12:07] Nebadon Izumi: i have never had that problem myself compiling mono
[12:07] Richardus Raymaker: and for some reason mono dont compile the 32bit version on 64bit system
[12:07] Nebadon Izumi: that does not suprrise me
[12:07] Richardus Raymaker: well. it only happend on suse 12.3 sofar
Personal tools
General
About This Wiki