Chat log from the meeting on 2018-01-09

From OpenSimulator

Jump to: navigation, search

TL;DR:
- Rezzing at Hurlington Plaza
- Progress of mute-list and estate-bandwidth
- issues with OS due to ancient releases of .net-frameworks, mono, glibc, kernel ...
- OS code barely compatible with .net coding style guidelines
- memory leaks in R.O.B.U.S.T
- Hints for MONO_ENV variables @11:53
- "unable to verify" message on an HG TP likely to be caused by stale DNS entries due to excessive caching
- Hints for integrating the mute-addon by Andrew

[10:51] George.Equus @hg.osgrid.org: Wow, fast rez tonight
[10:52] Sheera Khan: well, sortof ;-)
[10:52] George.Equus @hg.osgrid.org: I got everything more or less instantly on screen
[10:52] George.Equus @hg.osgrid.org: Unusual
[10:53] Sheera Khan: ahhh, another prim appeared ^^
[10:53] Sheera Khan: ohhh, and a second one
[10:54] George.Equus @hg.osgrid.org: Directors chairs don't render properly though for me, ned to cam forth and back to see legs
[10:54] Sheera Khan: at this rate the region will be mostly be loaded by the end of the meeting
[10:55] George.Equus @hg.osgrid.org: Poor bandwidth?
[10:55] Sheera Khan: hmmm, wouldn't say so... it's 50MBps down and 10 MBps up
[10:56] Sheera Khan: uiiii, the chairs rezzed in ^^
[10:56] Sheera Khan: ping time is good for intercontinental... at about 150ms
[10:57] George.Equus @hg.osgrid.org: Bandwidth certainly is enough for you. First time I went here after big yearly clean-out (emptying cache among other things...) it took forever to load, after that only getting better each time.
[10:57] Leighton.Marjoram @grid.opensim.life:8002: waves hello to all :)
[10:58] George.Equus @hg.osgrid.org: Hi Leighton
[10:58] Selby.Evans @grid.kitely.com:8002: hi Leighton
[10:58] Sheera Khan: ok, i cleared the cache bc I'm on the FS beta now
[10:58] George.Equus @hg.osgrid.org: my ping is steady around 100
[10:59] Sheera Khan: I guess this server is in the US and I'm in germany... so 150ms is fine :-)
[11:00] George.Equus @hg.osgrid.org: Yes ... in Sweden myself
[11:00] Andrew.Hellershanks @hg.osgrid.org: Net distance and geographic distance aren't related.
[11:00] Andrew.Hellershanks @hg.osgrid.org: Not that closely at any rate.
[11:01] Andrew.Hellershanks @hg.osgrid.org: George, run a traceroute.
[11:01] Ramesh.Origin @login.rezmela.net:8002: just fixing my clothes
[11:01] Sheera Khan: well server in Germany have about 30ms ping time for me
[11:04] George.Equus @hg.osgrid.org: Getting around 35 ping from German server where Maritime is now
[11:07] Andrew.Hellershanks @hg.osgrid.org: I was just checking the latest changes to the code. Understandably development has been on the quiet side over the holidays.
[11:07] Andrew.Hellershanks @hg.osgrid.org: I'm going to put Ubit in the spotlight for a moment.
[11:08] Sheera Khan: the mute module seems to have made progress though :-)
[11:08] Andrew.Hellershanks @hg.osgrid.org: Ubit, I see there were some changes re: mutes. Have you finished the work on your mutelist module or is there more work still pending?
[11:08] Ubit Umarov: hmm me? spot light ?
[11:08] Andrew.Hellershanks @hg.osgrid.org: Like you haven't been there before. :)
[11:09] James.atLLOUD @hg.osgrid.org: @leighton chat window - lol -I know, happens to me all the time. Wish Firestorm woulda delivered that modal(?) window version of the viewer.
[11:09] Ubit Umarov: at least, the module needs to be made the default ( rename, remove xmutes... )
[11:09] Andrew.Hellershanks @hg.osgrid.org nods
[11:09] Andrew.Hellershanks @hg.osgrid.org: We won't need XMutes once your module is ready for use.
[11:09] Ubit Umarov: for now to easy testing it is still named MutesModuleTst
[11:10] Bill.Blight @grid.opensim.life:8002: Also , Ubit created a work around for the non functional region ban lists
[11:10] Andrew.Hellershanks @hg.osgrid.org: I haven't tried (or needed) banning anyone so I wasn't aware there was an issue with ban lists.
[11:11] Bill.Blight @grid.opensim.life:8002: Well needing them and the fact they didn't work are two diff issues, but yes they were broken
[11:11] Ubit Umarov: this is about estate ban list
[11:12] Bill.Blight @grid.opensim.life:8002: yes sorry I meant region/estate ban lists
[11:12] Andrew.Hellershanks @hg.osgrid.org: True, Bill.
[11:12] George.Equus @hg.osgrid.org: Only banned compulsory Friend invites on some regions I go to often
[11:12] Ubit Umarov: think they never worked before, bc the comms protocol wasn't just there ( or incomplete)
[11:12] Andrew.Hellershanks @hg.osgrid.org: Has the work around been pushed? I don't see a reference to it in git log.
[11:13] Bill.Blight @grid.opensim.life:8002: http://opensimulator.org/viewgit/?a=commit&p=opensim&h=803289877307ce0016627f58af382b68a1905327
[11:13] Andrew.Hellershanks @hg.osgrid.org: ty, Bill. I see it now. It was regarding mantis 8271.
[11:14] Andrew.Hellershanks @hg.osgrid.org: Sounds good, Ubit. A lot of grid users will be happy to know that is working for when they move to 0.9
[11:16] Andrew.Hellershanks @hg.osgrid.org: I don't have last weeks meeting notes in front of me. Someone was going to be checking in to an issue this past week. I can't remember who it was or what was the issue.
[11:17] Andrew.Hellershanks @hg.osgrid.org: No one is speaking up so it may be someone at last weeks meeting who isn't here today.
[11:18] Andrew.Hellershanks @hg.osgrid.org: Any one have any issues to discuss?
[11:19] Ubit Umarov: oh i already removed xmutes on master ?
[11:19] Andrew.Hellershanks @hg.osgrid.org: Any newly discovered problems?
[11:19] James.atLLOUD @hg.osgrid.org: IIRC last weeks notes seemed to be mostly about performance.
[11:19] Ubit Umarov: well im having a big issue now
[11:19] Ubit Umarov: minimal .net framework suport
[11:20] Bill.Blight @grid.opensim.life:8002: I know there was some discussion about mesh loading being slow last week, but I don't seem to notice it ..
[11:20] Ubit Umarov: ( and so minimal mono.. minimsl linux kernel.. glibc.. win version )
[11:20] George.Equus @hg.osgrid.org: On a side note: Windows 8 support ends today
[11:20] Ubit Umarov: thing is that is being harder to keep 4.0 as target/minimal version
[11:20] Bill.Blight @grid.opensim.life:8002: Yes that glibc , we are stuck in the stone ages just to support OLDER linux and CentOS
[11:21] Bill.Blight @grid.opensim.life:8002: OLDER = Ancient in linux terms
[11:21] Ubit Umarov: problem is that a framework is a lot of things.. compiler version and language features
[11:21] Ubit Umarov: plus a ton of libraries
[11:21] Ubit Umarov: opensim of course does use a lot of those libraries
[11:22] Andrew.Hellershanks @hg.osgrid.org: Which .net framework to support has been a topic of discussion yet to be resolved.
[11:22] Ubit Umarov: and happens they do have bugs, that only get fixed on newer versions
[11:22] Bill.Blight @grid.opensim.life:8002: 4.5+ and even 4.0 on some systems
[11:22] Ubit Umarov: and some of those bugs are nasty security issues
[11:22] Bill.Blight @grid.opensim.life:8002: What I meant was even 4.0 has issues compiling on some systems
[11:23] Ubit Umarov: one in particular was only fixed on 4.5.2 ( windows who knows on mono )
[11:23] Sheera Khan: CentOS 7.3 uses glibc 2.17
[11:23] Andrew.Hellershanks @hg.osgrid.org: If (when?) the .net version required gets bumped who will that impact?
[11:23] Ubit Umarov: even 4.5 in testing on httptests is now obsolete
[11:24] Ubit Umarov: in windows world, XP is being deleted from even all references.. it is not suported, can only instal up to .net4.0
[11:24] Bill.Blight @grid.opensim.life:8002: Ubunu 17.10 uses glibc 2.26
[11:25] Andrew.Hellershanks @hg.osgrid.org: I just recently updated to Mint 18.3 and I have libc 2.23
[11:25] Sheera Khan: and you're right, that's pretty ancient ... ~ 2012^^
[11:25] Ubit Umarov: windoes vista is still listed and can install up to .net 4.6
[11:25] Bill.Blight @grid.opensim.life:8002: I think the latest mono does do up to .net 4.6 if I remember correctly
[11:26] Bill.Blight @grid.opensim.life:8002: but getting it to compile is another story
[11:26] Ubit Umarov: ao at this point im starting to think a jump to 4.6 is needed.. or at least 4.5.2
[11:26] Andrew.Hellershanks @hg.osgrid.org: Bill, which version of mono?
[11:26] Bill.Blight @grid.opensim.life:8002: whatever the newest is, I'll get the version, just a sec
[11:27] Andrew.Hellershanks @hg.osgrid.org: ok.
[11:27] Bill.Blight @grid.opensim.life:8002: 5.4.1.6
[11:27] Ubit Umarov: ( this in windows versions. respective mono ones is very badly docomented by mono )
[11:27] Andrew.Hellershanks @hg.osgrid.org: I haven't updated my mono in a while. I might even be still using 3.2.8 on my machine.
[11:27] Andrew.Hellershanks @hg.osgrid.org: They are in to 5.x now? Doesn't seem long since mono got up to 4.4 or 4.6
[11:28] Ubit Umarov: so should we totally give up XP ( and linux of same age ? )
[11:28] Bill.Blight @grid.opensim.life:8002: from the mono 5x release notes .. the reference assemblies were updated to fully match the .NET 4.6.2 API set.
[11:28] Andrew.Hellershanks @hg.osgrid.org: I still have XP in virtualbox but not for anything that matters.
[11:29] Ubit Umarov: yeap and it fails compiling that .net40 property Bill.Blight
[11:29] Bill.Blight @grid.opensim.life:8002: yes
[11:29] Bill.Blight @grid.opensim.life:8002: but pushing the target framework to 4_5 it does compile, at least on my box
[11:29] Ubit Umarov: still present on .net4.7.1
[11:29] Andrew.Hellershanks @hg.osgrid.org: When I was trying to compile a mono 4.x (or was it only 3.2.8) I had to bootstrap by building and installing 2.10 and working my way up.
[11:30] Ubit Umarov: well my code on it can be removed on 4.6+ ufff
[11:30] Ubit Umarov: 4.5.2 made that i was forced to add the default
[11:30] Bill.Blight @grid.opensim.life:8002: It seems there is either going to be some unhappy campers, or opensim stays stuck in the past
[11:31] Ubit Umarov: i would like keeping XP... but seeing more and more issues on that
[11:31] Ubit Umarov: bc ms and others are very active killing it
[11:32] Andrew.Hellershanks @hg.osgrid.org: At some point we will have to update. We need to make sure that a bump won't require an unstable/unusable version of mono.
[11:32] Bill.Blight @grid.opensim.life:8002: The same people who want Opensim to move forward, don't want to actually move forward, at least some of the vocal ones ... LOL
[11:32] Andrew.Hellershanks @hg.osgrid.org: Remember the problems we had with the early 4.x versions and in particular 4.2
[11:32] Ubit Umarov: note that such a update is not without issues
[11:32] Ubit Umarov: somethings may get broken and hard to debug
[11:32] James.atLLOUD @hg.osgrid.org: @bill - too true
[11:32] Andrew.Hellershanks @hg.osgrid.org: Very true, Bill. There are a lot of users out there who are still holding back on the move to 0.9
[11:33] Ubit Umarov: because the so claimed backwards compatibilty ms did promise is lost
[11:33] Ubit Umarov: new frameworks versions do break code for old ones
[11:33] Bill.Blight @grid.opensim.life:8002: Well there is backwards compatibility and then there is Legacy , they dropped legacy, you can't go backwards forever ...
[11:34] Leighton.Marjoram @grid.opensim.life:8002: nods @ Bill
[11:34] Andrew.Hellershanks @hg.osgrid.org: If the required framework gets bumped to, say, .net 4.5, what does that mean in terms of possible changes to the OS code?
[11:34] Ubit Umarov: well 4.5.2 breaking 4.0 is well..
[11:34] Ubit Umarov: there is a list of some known issues
[11:35] Ubit Umarov: but out code is 2.0 in someplaces
[11:35] Bill.Blight @grid.opensim.life:8002: yep, that is a issue would mean a separate branch for those unwilling to move forward
[11:35] Sheera Khan: would it be a possible solution to provide suitable mono versions for the mainstream distributions along with opensim?
[11:36] Bill.Blight @grid.opensim.life:8002: We recently saw this issue, when core tried to update the Jenkins box
[11:36] Ubit Umarov: the use of "clever" features increases the odds of breaking on upgrades
[11:36] Bill.Blight @grid.opensim.life:8002: just way too out of date in the OS itself
[11:36] Andrew.Hellershanks @hg.osgrid.org: Bill, The code might be branched in that case but it could also mean users on that branch of older code would not see much in the way of any ongoing development.
[11:36] Bill.Blight @grid.opensim.life:8002: yep
[11:37] Bill.Blight @grid.opensim.life:8002: and that is when the crying starts
[11:37] Ubit Umarov: i did test a compatibility analyses code
[11:37] Ubit Umarov: it crashed :p
[11:37] Andrew.Hellershanks @hg.osgrid.org: oops :)
[11:37] Andrew.Hellershanks @hg.osgrid.org: Not a good start
[11:37] Bill.Blight @grid.opensim.life:8002: I ran the OS code through one of those once, I swear it laughed at me
[11:37] Sheera Khan: let's call it OpenSim-legacy and the new ones OpenSim-Cool ^^
[11:37] Ubit Umarov: using stock code analyses i only see come 4500 warnings :p
[11:37] Leighton.Marjoram @grid.opensim.life:8002: hehehe Sheera
[11:38] George.Equus @hg.osgrid.org: 4.5.51650 latest (last?) installed on my win 8.1 ....
[11:38] Bill.Blight @grid.opensim.life:8002: Opensim-Cool, where the cool kids hang
[11:38] Ubit Umarov: many bc our so called events everywhere violate ms code rules
[11:38] Andrew.Hellershanks @hg.osgrid.org: Oh, is that all? You'll have those fixed in a week or two, Ubit. ;)
[11:38] Ubit Umarov: other issues are abusive use ov virtual functions with circular references
[11:39] Bill.Blight @grid.opensim.life:8002: 4500 warnings that is like 47 trips to the coffee shop for Ubit
[11:39] Andrew.Hellershanks @hg.osgrid.org: Something is allowing circular references? That sounds like a sure recipe for disaster.
[11:39] Sheera Khan hands Ubit a big mug of Java
[11:39] Bill.Blight @grid.opensim.life:8002: there are circular references all over the OS code
[11:40] Bill.Blight @grid.opensim.life:8002: hard to NOT find one
[11:40] Ubit Umarov: some are funny i did worked on then changing to the recomended code pattern.. doing get got a warning telling to use the code pattern i had before :)
[11:40] Andrew.Hellershanks @hg.osgrid.org: Bill, I don't look that closely at the code. :)
[11:41] Bill.Blight @grid.opensim.life:8002: LOL Andrew that is not a good admission for a member of core
[11:41] Bill.Blight @grid.opensim.life:8002: LOL
[11:41] Bill.Blight @grid.opensim.life:8002: I mean we know that, but don't admit it ... HAHAHA
[11:41] Andrew.Hellershanks @hg.osgrid.org: Bill, not really. I've made it clear before I don't do a lot of coding for the project.
[11:42] Bill.Blight @grid.opensim.life:8002: Just yanking your chain andrew
[11:42] Andrew.Hellershanks @hg.osgrid.org: I stick to supporting addon modules and certain other issues that don't require me to dig too deep in to the code.
[11:42] Andrew.Hellershanks @hg.osgrid.org: Bill, :)
[11:42] Andrew.Hellershanks @hg.osgrid.org: I have found monodevelop has made it easier to find my way around in the code.
[11:43] James.atLLOUD @hg.osgrid.org: ubit is there a tool you used to test compatibility?
[11:44] Ubit Umarov: i tried one.. it crashed
[11:44] Andrew.Hellershanks @hg.osgrid.org: I'd like to find a program like valgrind that would work with C#
[11:44] Ubit Umarov: both on vs2017 and 15
[11:44] James.atLLOUD @hg.osgrid.org: ok, so the warnings came from a test install I guess. interesting.
[11:45] Bill.Blight @grid.opensim.life:8002: I forget the one I used now, but when I ran it through on VS it just puked, and flipped me off
[11:45] James.atLLOUD @hg.osgrid.org: <googling valgrind>
[11:45] Ubit Umarov: err the warnings are for code analysis.. a bit diferent
[11:45] Ubit Umarov: ( but can mean issues )
[11:45] James.atLLOUD @hg.osgrid.org: Ah, ok. thanks.
[11:46] Andrew.Hellershanks @hg.osgrid.org: James, valgrind is a very useful tool for doing memory use analysis. Helps you find memory leaks in a (usually C based) program
[11:46] James.atLLOUD @hg.osgrid.org: memory leak = circular calls?
[11:46] Andrew.Hellershanks @hg.osgrid.org: I'm still of the opinion there is a memory leak in Robust but I don't have a tool that would help me track it down.
[11:46] Bill.Blight @grid.opensim.life:8002: what someone should do, NOT VOLUNTEERING, is also run it through mono profiling to find where the leaks and such are in mono
[11:46] Andrew.Hellershanks @hg.osgrid.org: James, allocating memory that never gets freed.
[11:47] James.atLLOUD @hg.osgrid.org: ok thank you.
[11:47] Andrew.Hellershanks @hg.osgrid.org: James, if a program keeps allocating new memory without properly freeing what is no longer in use the program will run out of available system memory and eventually crash.
[11:48] James.atLLOUD @hg.osgrid.org: I remember those in Windows - has got to be really difficult to diagnose.
[11:49] Andrew.Hellershanks @hg.osgrid.org: James, it can be difficult. The right tools can help to narrow down the problem.
[11:49] Ubit Umarov: possible httptests is doing better on that
[11:49] Ubit Umarov: at least acording to bill.Blight
[11:49] Ubit Umarov: ( robust mem use )
[11:49] Bill.Blight @grid.opensim.life:8002: yes it is
[11:50] James atLLOUD: thanks - sorry to interrupt but I'm grateful for the explains :)
[11:50] Andrew Hellershanks: I will be updating a grid to latest git master soon so I will be checking if there are any changes that have impacted the use of memory by Robust.
[11:50] Ubit Umarov: mem use is a pain for us
[11:50] Andrew Hellershanks: James, np.
[11:50] Bill.Blight @grid.opensim.life:8002: I don't have my stats in front of me, but I'll try and remember to bring some to the next meeting
[11:50] Andrew Hellershanks: Indeed. I've had to install a cron job to restart Robust 3 times a day to prevent it from eating up all system RAM.
[11:51] Ubit Umarov: some cases is pure leaks. we are not releasing resources... thats ok hard to find but fixable
[11:51] Bill.Blight @grid.opensim.life:8002: but I don't see the ever expanding memory usage on robust on the httptests branch
[11:51] Ubit Umarov: other issue is GC
[11:51] Ubit Umarov: GC does not release resources in same cases
[11:51] Ubit Umarov: and we can't do anything on that
[11:51] Bill.Blight @grid.opensim.life:8002: but also using the mono ENV switches that Ubit and I beat to death
[11:52] Andrew Hellershanks: Ubit, if memory was released new calls should either trigger a GC or it should issue memory that had been recently freed.
[11:52] Ubit Umarov: yeap changing the gc mode and defualts does help
[11:52] Ubit Umarov: GC as assuptions about objects life
[11:52] Andrew Hellershanks: Ubit, I'll have to talk with you about that later on IRC. I'd like to check what other grid config changes I can make that may help matters.
[11:52] Ubit Umarov: some it just decides are ehternal and removes them from any checks
[11:53] Ubit Umarov: etc
[11:53] Bill.Blight @grid.opensim.life:8002: Incoming chat spam
[11:53] Bill.Blight @grid.opensim.life:8002: export MONO_GC_PARAMS="nursery-size=32m,promotion-age=14,minor=split,major=marksweep,no-lazy-sweep,alloc-ratio=50"
export MONO_GC_DEBUG=""
export MONO_ENV_OPTIONS="--desktop"
export MONO_THREADS_PER_CPU=250
[11:53] Andrew Hellershanks: Bill, btw, mono profiling can generate a lot of output and that can take a while to sort through and understand
[11:53] Ubit Umarov: gc options? thing Bill did test even more than i did
[11:53] Ubit Umarov: note those req a recent mono
[11:53] Andrew Hellershanks: ty, Bill. I'll check on that.
[11:54] Andrew Hellershanks: Ubit, how recent?
[11:54] Ubit Umarov: 4.6.?? i guess
[11:54] Bill.Blight @grid.opensim.life:8002: 4.6+
[11:54] Bill.Blight @grid.opensim.life:8002: yes
[11:55] Bill.Blight @grid.opensim.life:8002: the --desktop switch makes the most difference in the GC as it assumes the GC should run more often, at least that is how it appears
[11:55] Andrew Hellershanks: ok. I have git master of mono checked out on my machine. I'll have to switch to the 4.6 branch
[11:55] Andrew Hellershanks: Not surprisingly, the more logins/logouts in a grid the faster Robust seems to use up RAM.
[11:56] Andrew Hellershanks: One quick question that I'm sure someone can answer before we hit the top of the hour.
[11:56] Andrew Hellershanks: I had a user ask about the "unable to verify" message you get on an HG TP to a new grid.
[11:57] Andrew Hellershanks: I know it was discussed on IRC but it was talked about a while ago and my long term memory isn't as long as it used to be. :)
[11:57] George Equus: Andrew, get that a lot lately too...
[11:57] Bill.Blight @grid.opensim.life:8002: that is generally cached DNS or mixing of ip and names on regions and in roubst
[11:57] SingerGirl.Mode @grid.opensim.life:8002: I've gotten it too.
[11:57] Andrew Hellershanks: oh, Right. Something related to reverse IP lookup??
[11:57] Bill.Blight @grid.opensim.life:8002: yes
[11:57] Bill.Blight @grid.opensim.life:8002: cached rDNS
[11:57] Ramesh.Origin @login.rezmela.net:8002: Thanks for asking the question Andrew
[11:57] Ubit Umarov flüstert: i dont on my 1 user grid :p
[11:57] Andrew Hellershanks: yea, I vaguely recall the issue.
[11:58] Bill.Blight @grid.opensim.life:8002: but cached forward can cause it too
[11:58] Andrew Hellershanks: Oh, right. Cached DNS data having out of date information.
[11:58] Bill.Blight @grid.opensim.life:8002: if someone has recently changed their IP
[11:58] George Equus: sometimes a renewed request for tp get me there eventually, but not always
[11:58] Bill.Blight @grid.opensim.life:8002: it is most prevalent on dynamic DNS "grids"
[11:58] Ubit Umarov flüstert: google public dns has issues with short ttl entries.. it refuses them
[11:59] Bill.Blight @grid.opensim.life:8002: I use the term "grid" very lightly when using it with dynamic DNS
[11:59] Andrew Hellershanks: How serious is the issue? Is a user going to run in to problems if they encounter that type of problem?
[11:59] Ubit Umarov flüstert: you will notice tps fail, you retry.. fail.. retry works
[11:59] Sheera.Khan @hypergrid.org:8002: well, they can't get to the destination ...
[11:59] George Equus: Right Ubit
[11:59] Ubit Umarov flüstert: happens a bit to me here at osg
[11:59] Bill.Blight @grid.opensim.life:8002: tell them to change their dns servers, or if it is cached on the "grid" side they need a shutdown/clear dns/restart
[12:00] Andrew Hellershanks: Ubit, definitely sounds annoying but as long as that is the main problem as a result.
[12:00] Bill.Blight @grid.opensim.life:8002: or if on linux kill the resolver service
[12:00] Andrew Hellershanks: How long does the TTL value need to be as a minimum for Google to accept it?
[12:00] Ubit Umarov: well i had that issue on the machine bill gave me to test
[12:00] Ubit Umarov: i change to another public dns services and worked
[12:00] Bill.Blight @grid.opensim.life:8002: we managed to kick it in the ass for a while, then Ubit burnt out the power supply ... HEHEHE
[12:01] Andrew Hellershanks: Bill :)
[12:01] Ubit Umarov: but ubuntu dns resolver also has issues.. i had to kill it on the machine from Bill.Blight
[12:01] Ubit Umarov: debian one on my recovered old pc seems ok
[12:02] Bill.Blight @grid.opensim.life:8002: Ubuntu and a couple other linux flavors use a caching DNS service, and it can cause the issue
[12:02] Ubit Umarov: oops it is down and is not rebooting lol
[12:02] Sheera.Khan @hypergrid.org:8002: systemctl disable nscd
[12:02] Bill.Blight @grid.opensim.life:8002: The best fix ... Quit trying to run a public grid on dynamic dns
[12:03] Ubit Umarov: "recoved" seems no longer valid :)
[12:03] Andrew Hellershanks: hm... I'm not sure which DNS servers I'm using at home these days. The info is in my router. I'll have to log in and check. I know I was using Google DNS at one point when the local ISP DNS service was acting up.
[12:04] Andrew Hellershanks: I have avoided using dynDNS type services.
[12:04] Bill.Blight @grid.opensim.life:8002: grids should also use something like OpenDNS that supports a short ttl
[12:04] George Equus: when I started I had dynamic DNS, knew it changed when could not get to my own regions  :) so checked and edited regions.ini.
[12:04] George Equus: Now on Static.
[12:05] Andrew Hellershanks: oh, right OpenDNS. I've seen that before. I might have been using it at one point. Their two main IPs are something like 222 and 220, or something along that line.
[12:05] Bill.Blight @grid.opensim.life:8002: well if you have a good DNS provider and use names, and people use DNS providers that support short TTL then dynamic is useable but no where near an optimal solution, and not really Opensim's problem
[12:06] Andrew Hellershanks: Unless a grid is in testing mode the TTL value should not be short.
[12:06] Bill.Blight @grid.opensim.life:8002: I agree andrew, but explain that to all these "grids" that are run on dynamic DNS from Home PC's
[12:06] Andrew Hellershanks nods
[12:07] George Equus: And most do not allow Static.. for private use, not without an extra (dire) fee....
[12:07] Bill.Blight @grid.opensim.life:8002: I'm gonna be blunt, It is not Opensim's job to fix a problem created by a user, that refuses, to understand the complexity of how Opensimulator works
[12:07] Andrew Hellershanks: True. You often have to pay extra to get static IP.
[12:08] Bill.Blight @grid.opensim.life:8002: Opensimulator was never meant to be point and click
[12:08] George Equus: My earlier provider charged hefty for it
[12:08] Andrew Hellershanks: Well said, Bill.
[12:08] George Equus: current one offer it for free
[12:08] Andrew Hellershanks: Running OpenSimulator in grid mode is not simple.
[12:08] Bill.Blight @grid.opensim.life:8002: and IMHO should not ever be
[12:09] Andrew Hellershanks: There are some grid owners who may wish it was so they wouldn't get as many headaches. :)
[12:09] Bill.Blight @grid.opensim.life:8002: The learning curve is just right for people who actually want to learn how to do it
[12:09] Bill.Blight @grid.opensim.life:8002: that is a double edged sword Andrew
[12:09] Bill.Blight @grid.opensim.life:8002: If they learned in the first place, they would not have as many headaches
[12:09] George Equus: took me two months to set it up lol, sure wasn't easy... knew zilch about networking at the time
[12:10] Andrew Hellershanks: There are a number of settings in the ini files that are a bit more obscure in terms of how they should be, or could be, set to optimize performance.
[12:10] Bill.Blight @grid.opensim.life:8002: and they are no where near optimized by default
[12:10] Bill.Blight @grid.opensim.life:8002: LOL
[12:11] Andrew Hellershanks: I've gotten grids running within a week or so but I still often wonder if there are other ini changes I should make to optimize things for best performance.
[12:11] Bill.Blight @grid.opensim.life:8002: If you don't understand how to backup and recover databases, don't run opensim
[12:11] George Equus: What Dan post is linux optimized, had to do some small changes to fit in nicely with Windows
[12:11] Bill.Blight @grid.opensim.life:8002: if you don't understand networking, don't run opensim
[12:11] Bill.Blight @grid.opensim.life:8002: if you don't understand file systems , don't run opensim
[12:11] Andrew Hellershanks: Sharing information on tweaking settings for optimization can help but that info also needs to be easily discoverable.
[12:11] Bill.Blight @grid.opensim.life:8002: if you don't understand how to read an ini , don't run opensim
[12:11] Bill.Blight @grid.opensim.life:8002: Until you at least learn those things
[12:12] Sheera.Khan @hypergrid.org:8002: I don't think everyone should host his/her own grid... If they don't want to learn how to do it properly they can use the services of one of the open grids...
[12:12] Bill.Blight @grid.opensim.life:8002: I agree sheera, there are too many tools that let people appear at first that they are a "grid runner" , until the first problem
[12:12] Andrew Hellershanks: Right. There are a lot of settings that need to be changed in the ini files when setting up a grid. If you can't read and understand the ini files you will have problems.
[12:13] George Equus: Don't think most users are that interested in running their own setup anyway, plenty of options around for those anyway
[12:13] Bill.Blight @grid.opensim.life:8002: ummm george , cough "outworldz", cough ...
[12:13] Andrew Hellershanks: People will soon learn that just because they *can* set up a "grid", doesn't mean they know how to run one. Running a grid is a lot of work.
[12:13] Andrew Hellershanks grins at Bill
[12:14] George Equus: :)
[12:14] Bill.Blight @grid.opensim.life:8002: I have 6 servers that run my grid, and I manage them all myself, with redundant backups mirror mysql , but you can run it as good on your desktop right?
[12:14] Ubit Umarov: sheera did u ppl tested the robust mutes ?
[12:14] George Equus: outworldz.... where did I hear about that.. hmmmm
[12:14] Andrew Hellershanks: Having good backups is very important. Accidents (and HD failures) happen. Avination is a good example. They had good backups and would have been online again after they had finished recovering their DBs from the backups.
[12:15] Andrew Hellershanks: Bill, I have run OS in grid mode on my desktop for testing with just a couple of regions.
[12:16] Bill.Blight @grid.opensim.life:8002: well running it for testing purposes is one thing, running and being the caretaker for others assets and their "homes" is an entirely different matter
[12:16] George Equus: I know far too little to manage anything but myself...
[12:16] Sheera.Khan @hypergrid.org:8002: @Ubit: not in our "production" grid as it still runs on 0.8+ :-( I was a bit out of the loop due to RL work lately so I'm not sure if it is already tested in our test grid
[12:16] Andrew Hellershanks: I look after one grid that has 6 servers.
[12:16] Andrew Hellershanks: Two are mostly used for backups.
[12:18] Andrew Hellershanks: The other issue with making OS "too easy" for people to use to set up their own grid is it just further fragments the userbase. Getting and maintaining users in a grid is another issue all grid owners face.
[12:19] Bill.Blight @grid.opensim.life:8002: well that has already happened
[12:19] Andrew Hellershanks: yes. It has been a problem for quite some time.
[12:19] Bill.Blight @grid.opensim.life:8002: now you have 1400 people staying at home on their "mini-grids" as they are called , rather than being part of the community
[12:20] Andrew Hellershanks nods
[12:20] Bill.Blight @grid.opensim.life:8002: those people venture out on the HG to get STUFF then go home
[12:21] Sheera.Khan @hypergrid.org:8002: and complain about nobody visiting them and OS being empty ^^
[12:21] Bill.Blight @grid.opensim.life:8002: yep
[12:21] Andrew Hellershanks: indeed
[12:21] Bill.Blight @grid.opensim.life:8002: and then complain because opensimulator does not do what THEY want
[12:21] Bill.Blight @grid.opensim.life:8002: they should just be able to click a button and make it do what they want
[12:22] Andrew Hellershanks: Right. Those people don't want to invest the time to learn about OS, what it can do, and how to customize it.
[12:22] George Equus: Don't like visitors unannounced... mean I would have to hoover... and remove redundant prims
[12:22] Bill.Blight @grid.opensim.life:8002: on that note, I am going to add that Opensimulator needs better access control settings .... <looks around all innocent> not that I'd want to ban anybody
[12:23] Andrew Hellershanks: OS is meant to be a starting point, not a final end product.
[12:23] Andrew Hellershanks: We are well past the hour now. Any other items before we wrap up todays meeting?
[12:23] Bill.Blight @grid.opensim.life:8002: yes , why I do my fork which is actually more of a "compliment" than a fork ..
[12:24] Andrew Hellershanks: I've been looking at some add on modules (mute for one) but before that was a birds module, and there is a destinations guide that I'm looking at.
[12:24] Bill.Blight @grid.opensim.life:8002: djphil has a decent destinations guide,
[12:24] Bill.Blight @grid.opensim.life:8002: jopensim now has one built into it
[12:25] Andrew Hellershanks: Bill, yes. That is the one I'm looking at. I want to finish a code review before deploying it.
[12:25] Sheera.Khan @hypergrid.org:8002: a better search engine would be a great thing too
[12:25] Bill.Blight @grid.opensim.life:8002: I'm going to be rolling your mutelist into my build andrew
[12:25] Bill.Blight @grid.opensim.life:8002: just getting the time to make sure it is seamless on the build process
[12:25] Andrew Hellershanks: Early versions has possible SQL injection issues. I think he has fixed those issues now but I still want to review the module code and the LSL scripts it uses.
[12:26] Bill.Blight @grid.opensim.life:8002: there is no module code for his
[12:26] Bill.Blight @grid.opensim.life:8002: it uses the code already in Opensim
[12:26] Bill.Blight @grid.opensim.life:8002: well
[12:26] Bill.Blight @grid.opensim.life:8002: actually it is all viewerside
[12:26] Bill.Blight @grid.opensim.life:8002: OS just provides the URL
[12:26] Andrew Hellershanks: Bill, You should not have to do anything more than copy the MuteListModule code in the sub-dir in to addons, run prebuild, and build as normal.
[12:26] Andrew Hellershanks: Then just put the PHP file in to place, and configuration.
[12:27] Bill.Blight @grid.opensim.life:8002: does it add the needed things to the SQL db, that appeared to be separate?
[12:27] Andrew Hellershanks: Bill, there is a .sql file you can run to set up the DB.
[12:27] Bill.Blight @grid.opensim.life:8002: right
[12:27] Bill.Blight @grid.opensim.life:8002: I want to include that in the migrations
[12:27] Bill.Blight @grid.opensim.life:8002: so it does not have to be added externally
[12:28] Andrew Hellershanks: Bill, ah, ok. You may have to make some changes to use it as part of a migration step.
[12:28] Bill.Blight @grid.opensim.life:8002: yep
[12:28] Bill.Blight @grid.opensim.life:8002: that is the only issue at the moment
[12:28] Bill.Blight @grid.opensim.life:8002: I've dug into a lot of OS code but not dug into that yet
[12:28] Bill.Blight @grid.opensim.life:8002: need to add an addon migration step
[12:29] Andrew Hellershanks: The migration thing isn't bad. You don't need to dig in to the code that much to see how its done. I've looked at that before.
[12:29] Andrew Hellershanks: The files are under OpenSim/Data/MySQL/Resources/ or some such.
[12:31] Andrew Hellershanks: You just add the steps needed to make the changes to the DB in to a file with a sort of marker line to know the steps for a given migration level.

Personal tools
About This Wiki