 Notes |
(0009023)
Teravus (administrator)
2009-01-26 01:55
|
Thanks for posting this.
I expect that this is a result of the images or a system interaction with openjpeg.
Please post relevant j2k streams and we'll send them off to the necessary people. The way the j2k decoder module is, if it encounters an error decoding the layers, it will simply queue up the full image to be sent. This means that eventually, it'll all come in, but it may take a while longer for it to then it would if the j2k decoder didn't run into an error. |
|
(0009027)
Teravus (administrator)
2009-01-26 03:08
|
setting to 'feedback' because we need more information from you to go any further with this bug report |
|
(0009236)
Snoopy (reporter)
2009-02-08 07:59
|
I experience the same problems under Ubunto 8.04 64 bits. All JPEG2000 texture decodings fail on my 64 bit server. See the following examples:
07:55:46 - [J2KDecoderModule]: JPEG2000 texture decoding failed for 0a48ae62-8597-4750-a554-d1f0b566266c
07:55:47 - Last packet end in JPEG2000 stream extends beyond the end of the file. filesize=120936 layerend=923107601
07:55:47 - [J2KDecoderModule]: JPEG2000 texture decoding failed for e8653b27-0c27-46c3-8774-b58ca0b023bb
07:55:47 - Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 138 End: -1023451028
Layer 1: Start: 589 End: -1023518944
Layer 2: Start: 1618 End: -1023597228
Layer 3: Start: 3577 End: -1023636572
Layer 4: Start: 9032 End: -1025005544
Layer 5: Start: 32257 End: -1026373012 |
|
(0009237)
Teravus (administrator)
2009-02-08 08:30
|
Please post relevant j2k streams and we'll send them off to the necessary people. The way the j2k decoder module is, if it encounters an error decoding the layers, it will simply queue up the full image to be sent. This means that eventually, it'll all come in, but it may take a while longer for it to then it would if the j2k decoder didn't run into an error. |
|
(0009242)
Snoopy (reporter)
2009-02-08 09:24
|
I get such error messages for all textures, without any exception. See the OSGrid UUIDs above. If you need more info please tell me how I can generate what you need. |
|
(0009247)
FredM (reporter)
2009-02-08 14:26
|
Same Problem here. Running on Debian unstable. Blurry images while texture (clothes) upload.
Console tells "JPEG2000 texture decoding failed for uuid....0129 and so on....
Now the strange thing: If I teleport to another region (same gridserver, same simulator) then ONE texture loads fine. Second texture blurry again!
I am runing on svn 8243 |
|
(0009249)
dmiles (reporter)
2009-02-08 14:38
|
This has been going on for several weeks with the svn trunk and still does with At revision 8287. |
|
(0009418)
dmiles (reporter)
2009-02-13 21:44
|
Steps To Reproduce:
Here are my system specs: Trying to be verbose here!
[root@titan bin]# gcc -v
Using built-in specs.
Target: x86_64-redhat-linux
Configured with: ../configure --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --with-bugurl=http://bugzilla.redhat.com/bugzilla [^] --enable-bootstrap --enable-shared --enable-threads=posix --enable-checking=release --with-system-zlib --enable-__cxa_atexit --disable-libunwind-exceptions --enable-languages=c,c++,objc,obj-c++,java,fortran,ada --enable-java-awt=gtk --disable-dssi --enable-plugin --with-java-home=/usr/lib/jvm/java-1.5.0-gcj-1.5.0.0/jre --enable-libgcj-multifile --enable-java-maintainer-mode --with-ecj-jar=/usr/share/java/eclipse-ecj.jar --disable-libjava-multilib --with-cpu=generic --build=x86_64-redhat-linux
Thread model: posix
gcc version 4.3.2 20081105 (Red Hat 4.3.2-7) (GCC)
Linux version 2.6.29-0.43.rc2.git1.fc11.x86_64 (mockbuild@x86-6.fedora.phx.redhat.com) (gcc version 4.3.2 20081105 (Red Hat 4.3.2-7) (GCC) ) 0000001 SMP Mon Jan 19 01:46:27 EST 2009
[root@titan mono]# mono -V
Mono JIT compiler version 2.3 (/trunk/mono r124332 Fri Jan 23 08:48:43 PST 2009)
Copyright (C) 2002-2008 Novell, Inc and Contributors. www.mono-project.com
TLS: __thread
GC: Included Boehm (with typed GC)
SIGSEGV: altstack
Notifications: epoll
Architecture: amd64
Disabled: none
FYI:
I tested it also with Mono 2.2 - Same thing
(Mono JIT compiler version 2.2 (/tags/mono-2-2/mono r126910 Fri Feb 13 17:48:45 PST 2009))
[root@titan mono]# uname -a
Linux titan 2.6.29-0.43.rc2.git1.fc11.x86_64 0000001 SMP Mon Jan 19 01:46:27 EST 2009 x86_64 x86_64 x86_64 GNU/Linux
Fresh checkout of OpenSim at revision 8386.
[root@titan bin]# pwd
/development/opensim/bin
[root@titan bin]# ls libope* -l
-rw-r--r-- 1 root root 125136 2009-02-13 21:36 libopenjpeg-dotnet-2.1.3.0-dotnet-1.dylib
-rw-r--r-- 1 root root 128100 2009-02-13 21:36 libopenjpeg-dotnet-2.1.3.0-dotnet-1.so
-rw-r--r-- 1 root root 129496 2009-02-13 21:36 libopenjpeg-dotnet-2.1.3.0-dotnet-1-x86_64.so
[root@titan bin]# ls ope* -l
-rw-r--r-- 1 root root 187392 2009-02-09 17:59 openjpeg-dotnet.dll
-rwxr-xr-x 1 root root 218 2009-02-09 17:59 opensim-ode.sh
Run opensim:
[root@titan bin]# mono OpenSim.exe
...
...
...
16:37:00 - [LOCAL INVENTORY SERVICE]: Received inventory response for user c04e7cb9-81f5-4726-b2c1-6850762bc1f0 containing 15 folders and 1 items
16:37:00 - [WORLD MAP]: Starting remote MapItem request thread
16:38:06 - [CACHE]: Caching d702e607-218d-466d-a6db-f98301e22ed2 for 24 hours from last access
16:38:06 - Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 208 End: 0
16:38:06 - [J2KDecoderModule]: JPEG2000 texture decoding failed for d702e607-218d-466d-a6db-f98301e22ed2
16:38:16 - [CACHE]: Caching 0c7398a2-7e4d-4766-80a1-0d0621e4f96d for 24 hours from last access
16:38:16 - Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 208 End: 0
16:38:16 - [J2KDecoderModule]: JPEG2000 texture decoding failed for 0c7398a2-7e4d-4766-80a1-0d0621e4f96d
16:38:19 - [CACHE]: Caching b7e2c845-c572-41ef-80a9-b16ec29a0fff for 24 hours from last access
16:38:20 - Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 208 End: 0
16:38:20 - [J2KDecoderModule]: JPEG2000 texture decoding failed for b7e2c845-c572-41ef-80a9-b16ec29a0fff
16:43:06 - [CACHE]: Caching 917ae9d3-b850-4c76-b3de-e971dbd84582 for 24 hours from last access
16:43:06 - Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 208 End: 0
This happens pretty consistantly.. on Linux 64
I cannot repo on Win32.
Sometimes win32 might say this though:
18:10:16 - [J2KDecoderModule]: JPEG2000 texture decoding threw an exception for
67b1c814-c24c-b6be-5f3d-1143a7022937, System.IndexOutOfRangeException: Index was
outside the bounds of the array.
However 95% of the time it give happy feedback on Win32
18:10:11 - [J2KDecoderModule]: 1969 Decode Time: e04e12ed-7ee9-4e56-93f6-f16f997
01dc5
18:10:12 - [J2KDecoderModule]: 250 Decode Time: 2d68244a-5b06-4192-ad40-b9e79136
f7ad
18:10:14 - [J2KDecoderModule]: 16 Decode Time: ed88284d-a566-4431-bb42-77815206b
16d |
|
(0009419)
dmiles (reporter)
2009-02-13 21:49
|
Addtional Info:
When I connect to the OpenSim server with LibOMV (TestClient)
Its constantly getting stuck 85-95% of texture asset downloads.. and upping the prioriy. (I suspect the server is never sending the the length correctly so th client gets stuck) |
|
(0009813)
FredM (reporter)
2009-02-28 04:29
edited on: 2009-02-28 04:45
|
I still get the same here, quite frustrating. (64B Bit Debian). I am on Mono 1.9.1, so I think it is maybe not related to Mono?
Last Opensim.log says this:
2009-02-28 13:17:24,322 WARN - OpenSim.Region.Environment.Modules.Agent.TextureSender.J2KDecoderModule [J2KDecoderModule]: JPEG2000 texture decoding failed for 9ded2189-5f6c-4d00-99d7-c20c80aab2dd
2009-02-28 13:17:24,334 ERROR - OpenMetaverse Inconsistent packet data in JPEG2000 stream:
and such messages:
2009-02-28 13:17:22,156 WARN - OpenSim.Region.Environment.Modules.Agent.TextureSender.J2KDecoderModule [J2KDecoderModule]: JPEG2000 texture decoding failed for c9c164ac-c5ba-499e-85c9-ddaec8c5027f
2009-02-28 13:17:22,397 ERROR - OpenMetaverse Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 135 End: 0
Layer 1: Start: 549 End: 0
Layer 2: Start: 1963 End: 0
Layer 3: Start: 3635 End: 0
Layer 4: Start: 9359 End: 7
Layer 5: Start: 32582 End: 0
Normal textures are blurry, some linked sculpties decay after awhile (well, they are organic :)) ) Single sculpties stay the way they are.
If I teleport out of the region and return, some of the textures finalise to load. Sculpies still a mess.
Restart of simulator no cure. No idea yet where to look else.
|
|
(0010141)
Jamenai (reporter)
2009-03-26 01:51
|
Same Problem in r8861 on 64bit openSuse 11.0 with mono 2.0 |
|
(0010156)
Ursula Matova (reporter)
2009-03-27 11:12
|
Sure it's an OpenJPEG / x64 issue ...
Still happens in svn.8899 ( mono 2.0.1 & 2.4 )
(( Linux Ubuntu 8.04 x86_64 )) |
|
(0010158)
jhurliman (manager)
2009-03-27 11:22
|
The proper solution would be to figure out why layer Start decodes fine but not layer End (people seem to be claiming it's a 64-bit issue). A hacky fix would be to treat properly decoded Start positions as a success and use layer[i + 1].Start -1 as the End parameter |
|
(0010196)
Teravus (administrator)
2009-03-28 11:34
|
The OpenJPEG has been updated with the LibOMV update yesterday. Can anyone confirm that they still have openJPEG decoding problems? |
|
(0010197)
Ursula Matova (reporter)
2009-03-28 11:44
edited on: 2009-03-28 11:45
|
Yes I can see some of these errors related to J2KDecoderModule.
But much less than before.
I'm using Mono 2.4 / Ubuntu 64Bits
|
|
(0010198)
Teravus (administrator)
2009-03-28 12:00
|
hmm, jhurliman: I could try the hacky fix, let you know. Ultimately there is still an issue somewhere in the decoding on some systems that renders 90% of the texture layer decodes useless. I'd jira it myself, however, I don't have a system where it repros |
|
(0010199)
dmiles (reporter)
2009-03-28 12:16
|
Are other reporters compiling mono --use-large-heap=yes?
I do.. I might break down and compile a normal mono. I just cannot comfortably run my software (including OpenSimular) w/o it and the GC patch at http://opensimulator.org/wiki/GC_NO_EXPLICIT [^] |
|
(0010200)
Teravus (administrator)
2009-03-28 12:34
|
The affected code is in libOMV, and not OpenSimulator. It looks like it will still complain even if I try to do some heuristics in the library consumer.
from LibOMV:
// More sanity checking
if (layerInfo.Length == 0 || layerInfo[layerInfo.Length - 1].End <= encoded.Length - 1)
{
success = true;
for (int i = 0; i < layerInfo.Length; i++)
{
if (layerInfo[i].Start >= layerInfo[i].End ||
(i > 0 && layerInfo[i].Start <= layerInfo[i - 1].End))
{
System.Text.StringBuilder output = new System.Text.StringBuilder(
"Inconsistent packet data in JPEG2000 stream:\n");
for (int j = 0; j < layerInfo.Length; j++)
output.AppendFormat("Layer {0}: Start: {1} End: {2}\n", j, layerInfo[j].Start, layerInfo[j].End);
Logger.Log(output.ToString(), Helpers.LogLevel.Error);
success = false;
break;
}
}
}
else
{
Logger.Log(String.Format(
"Last packet end in JPEG2000 stream extends beyond the end of the file. filesize={0} layerend={1}",
encoded.Length, layerInfo[layerInfo.Length - 1].End), Helpers.LogLevel.Warning);
} |
|
(0010201)
lmmz (reporter)
2009-03-28 13:19
|
Yeah, I'm still seeing the error in r8914. |
|
(0010202)
svn (developer)
2009-03-28 13:50
|
teravus committed related code in r8917
* Adding some heuristic error correction to the j2k decoder module to combat some of the situations that we see in mantis 3049 .
* This may help people on certain 64 bit systems where the end byte position of each layer data packet is incorrect but the start positions are correct.
* The console will still be extremely chatty with 'Inconsistent packet data in JPEG2000 stream:' messages, however.. if OpenSimulator was able to recover the data, it will say HURISTICS SUCCEEDED
see more at - http://opensimulator.org/cgi-bin/viewvc.cgi?view=rev&revision=8917 [^]
|
|
(0010203)
Teravus (administrator)
2009-03-28 14:52
edited on: 2009-03-28 15:02
|
Layers tested
Start|End|Size
Heuristics error recovery
- to -
Correct Decoding
138|591|453
592|1831|1239
1832|3405|1573
3406|9110|5704
9111|32140|23029
32141|130906|98765
- to -
138|595|457
596|1991|1395
1992|7996|6004
7997|31714|23717
31715|127532|95817
127533|130903|3370
132|577|445
578|1456|878
1457|2404|947
2405|24478|22073
- to -
132|580|448
581|1983|1402
1984|7759|5775
7760|24475|16715
138|593|455
594|1920|1326
1921|4319|2398
4320|9281|4961
9282|32495|23213
32496|130961|98465
- to -
138|597|459
598|1989|1391
1990|7945|5955
7946|31984|24038
31985|127677|95692
127678|130958|3280
So, the start packet byte offsets are, indeed, incorrect
|
|
(0010204)
lmmz (reporter)
2009-03-28 14:53
|
I'm now seeing "[J2KDecoderModule]: HEURISTICS SUCCEEDED ." :) |
|
(0010235)
Ursula Matova (reporter)
2009-03-30 10:22
|
Same, It's much much better :)
And, now, the "j2kDecodeCache" sub-folder is full of ".cache" files ...
Seems perfect for me ( SVN.8921 / Linux x64 ).
Thanks a lot Teravus ... |
|
(0010274)
Ursula Matova (reporter)
2009-03-31 12:11
|
Just upgraded to SVN.8946 ( Linux x86_64 / Mono 2.4 ).
Seems that everything working fine on my server with Teravus "Hacky Fix",
Any other feedback ? |
|
(0010275)
FredM (reporter)
2009-03-31 12:16
|
Yup, looks good to me as well! No errors in the console anymore!
The only problem still remaining are the blurry textures. Not sure I have to upload them again or it is just the Openjpg. |
|
(0010277)
Ursula Matova (reporter)
2009-03-31 12:25
|
I haven't noticed this "blurry textures" issue yet ... even with textures I have uploaded when this openJpeg was present.
May I suggest you to clear the following :
- Basicaly : your viewer cache ( even do it manually it's more efficient ),
- Delete "addin-db-001" drectory,
- Clean "j2kDecodeCache" directory.
Let's see if it solves your problem. :) |
|
(0010278)
FredM (reporter)
2009-03-31 13:08
|
Ursula,
thanks, good idea!
Viewer Cache I delete often - no difference
addin-db-001 I delete before each svn upgrade - no difference
j2kDecoderCache... I can try that out |
|
(0010279)
Ursula Matova (reporter)
2009-03-31 13:14
|
Just reproduced your "blurry textures" problem.
In SVN.8946, when you upload a texture, this one stay "blurry" ...
- Nothing relevant reported on server consoles,
1/ - Try to logoff/logon back : Still blurry texture,
--> If I select the texture on an object, then "Advanced Menu > Rendering > Selected Texture Info" report the folowing :
32x32 opaque on face 0
--> If I select another texture ( displaying fine ) :
512x512 opaque on face 1 2 3 4 5 6 7
2/ - Try to restart OpenSim server :
--> No change
3/ - Try to run "predecode-j2k" :
--> I have noticed the folowing messages :
22:07:10 - [J2KDecoderModule]: HEURISTICS SUCCEEDED 5 Decode Time: 00000000-0000-2222-3333-100000001007
22:07:11 - [J2KDecodeCache]: Cache Read failed. IO Exception.
22:07:11 - Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 135 End: 0
Layer 1: Start: 578 End: 0
Layer 2: Start: 1746 End: 0
Layer 3: Start: 2517 End: 0
Layer 4: Start: 8628 End: 0
4/- ReTry after cleaning the "j2kDecodeCache" directory :
--> Same errors as above ( about 5% of "J2KDecoderModule" are errors ).
AND : A LOT OF TEXTURES ARE BLURED :S |
|
(0010280)
FredM (reporter)
2009-03-31 13:50
|
PS: I updated from Mono 1.9.1 to 2.0.1 now. No difference. |
|
(0010281)
Ursula Matova (reporter)
2009-03-31 13:53
|
I'm using mono 2.4 ( haven't tried another mono release ).
I think this issue should be promoted to "Critical" ... :) |
|
(0010282)
Ursula Matova (reporter)
2009-03-31 14:27
|
I have also noticed many of the folowing errors on the AssetServer console :
22:06:41 - [REST]: GET:/asset failed to find 7cf259ee-5c63-9f1d-0d19-e07212535925
22:06:41 - [REST]: GET:/asset failed to find c0572bf5-eb50-580f-56f6-1a1ceb67a3ed
22:06:43 - [REST]: GET:/asset failed to find dd1a41a9-15d9-2870-b7f6-dd7d9b7842a8
22:06:44 - [REST]: GET:/asset failed to find 579cec25-22a2-22d0-06c3-f7faf8241990
Don't know if there is a link with our problem. |
|
(0010284)
Ursula Matova (reporter)
2009-03-31 15:05
|
Seems to be better with svn.8948 ... Almost all textures in scene display fine.
But seems that freshly uploaded textures keep blured.
- re-run "predecode-j2k" console command ... not helping at all.
- restarting the viewer doesn't help as well.
Last information : Displaying a texture from inventory also produce "blured" textures. |
|
(0010285)
Teravus (administrator)
2009-03-31 20:30
|
:( |
|
(0010288)
jhurliman (manager)
2009-03-31 21:48
|
Can someone upload a texture where the decode is failing and Teravus' heuristics kick in? A log of heuristics output for that texture would be ideal too, but not as important. |
|
(0010306)
Ursula Matova (reporter)
2009-04-01 10:22
|
I can upload a couple of textures, but it's not really interresting, because after a very very long time ( more than 5 hours ) the texture finaly display fine.
Note: SVN.8959 is now "J2KDecoderModule" quiet :) |
|
(0010307)
jhurliman (manager)
2009-04-01 10:40
|
That sounds more like a client issue (the client is responsible for requesting different quality layers of a texture). |
|
(0010308)
Ursula Matova (reporter)
2009-04-01 11:18
|
Maybe yes ...
I'm using Hippo OpenSim Viewer 0.5
Haven't tried with the regular S/L Viewer. |
|
(0010349)
Fenton Maurer (reporter)
2009-04-03 02:09
edited on: 2009-04-03 02:11
|
I continue to have messages and missing textures
OpenSuse 11.1 64bits
mono-core-2.0.1-1.19
nant-0.86_beta1-37.47
OpenSim trunk 8971
The viewer is Hippo 0 5.1 but also SL standard viewer 1.22.11 has same problem.
I see in log the HEURISTICS SUCCEEDED messages ... but after 1 hour not all the textures appears: land lacks squares (i see water squares in middle of the land) and trees are not shown.
I tried these trunk
8710 8942 8971
and all have the same problem.
|
|
(0010498)
Jamenai (reporter)
2009-04-06 20:17
edited on: 2009-04-06 20:18
|
### Testingresults:
# The Problem: Concerns following situations:
- Uploading new texture
-> Preview remains blurred
-> On Prim remains blurred
(long, long ... ... long time)(Format,size,tiled or not tiled ... simply ALL)
- Using a texture from the inventory
-> Preview remains blurred
-> On Prim remains blurred
(long, long ... ... long time)(Format,size,tiled or not tiled ... simply ALL)
# The System:
Suse Linux 64 bit (Rootserver like)
(Gridmode, Multiple Regions (6), Storeage: MySQL 5)
# System-Updates without effects
1. I updated from mono 1.9.1 -> 2.0.1 -> 2.4
-> Result: No difference
2. I updated and connected with SL standard Viewer and Hippo 0.4 / 0.5 / 0.5.1 -> Result: No difference
3. I updated to latest r9029
-> Result: No difference
# What also does not help:
- restarting the UGAIMS
- restarting the Client-Viewer/Client-PC
- reboot complete all
# My personal comment
Testing sucks without positive results oO
Any Ideas?
|
|
(0011424)
Fenton Maurer (reporter)
2009-05-12 11:30
|
Going on in testing .. the problem seems related to Udp packets corruption and/or loosing. Changing a router between my server and external internet, the problem is greatly reduced.
Now I see the J2KDecoderModule warning messages, but the textures are all loaded in relatively fast time (resend of udp packet is reduced).
Passed from a hour to load 4000 object to only 2 minutes ! |
|
(0011425)
Ursula Matova (reporter)
2009-05-12 12:09
|
Confirmed in my case ( svn.9477 ). For now, everything is working fine.
I think we should keep this mantis "open" to keep an eye on this texture/j2k issue. |
|
(0012103)
Fenton Maurer (reporter)
2009-06-11 02:20
|
Bad news! I acnnot have my server running until january. I contnue to pay for housing and bandwidth, but I'm stuck on this problem.
Now I moved to another dedicated server, with a lot of bandwith and housed at another datat center, no ADSL or routers around, the OS is Opensuse 64 10.1, Mono 2.0.1, 4GB Ram
I compiled OpenSIm 9843 from scratch and created a new region, to avoid any past error. the result was terrible! At least before I could login to my server!
The new server initializes and registers itself correctly to the grid, but if I try to login (Hippo), login stops at "waiting for region handshake" then "we're truoble connecting".
The log reports:
2009-06-11 11:13:18,357 DEBUG - OpenSim.Region.Framework.Scenes.Scene [SCENE]: Adding new child agent for Fenton Maurer in FMal_NEW
2009-06-11 11:13:18,471 WARN - OpenSim.Region.CoreModules.Agent.IPBan.SceneBanner [IPBAN] User "87.23.41.172" not in any ban lists. Allowing connection.
2009-06-11 11:13:18,472 DEBUG - OpenSim.Region.ClientStack.LindenUDP.LLClientView [CLIENT]: Entered main packet processing loop for Fenton Maurer in FMal_NEW
2009-06-11 11:13:21,468 ERROR - OpenMetaverse Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 138 End: 364
Layer 1: Start: 569 End: 160
Layer 2: Start: 1304 End: 168
Layer 3: Start: 3029 End: 72
Layer 4: Start: 9104 End: 224
Layer 5: Start: 32229 End: 336
and following a long series of these JPEG2000 errors.
Please ... help me, any suggestion is useful: I need to have a 64 bits os running, I followed all manuals and howto, but since January I cannot no more use OpenSim (before it has worked for months) |
|
(0012166)
Kenjay (reporter)
2009-06-16 14:32
|
I'm getting this now too, since upgrade from older level:
17:29:50 - [J2KDecoderModule]: JPEG2000 texture decoding failed for 90dd433f-ef88-cc4e-1f2b-eb245d5c5d90. Is this a texture? is it J2K?
17:29:50 - [ASSET TRANSACTIONS]: Uploaded asset data for transaction f1e39804-6c3a-c999-382a-633bab3fa06d
17:29:53 - Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 131 End: 0
It seems to only affect my hair and eyes, although changing to non-cached clothing also triggers it. Texttures go white. Environment is Mono on Linux, latest build, Server and client on same LAN, no internet. |
|
(0012400)
nixnerd (reporter)
2009-07-12 11:10
|
Not sure if this is related - but I'm running SVN 10019 on Debian Sid with mono 2.0.1 standalone using MySQL.
Certainly over quite a few recent SVN revisions, I've noticed that avatars eyes appear completely white, regardless of which textures are loaded for the avatars eyes - or wether new eyes are made and then worn - they all appear as white textures.
However - In appearance mode - all looks as it should. Rebaking makes no appreciable difference.
No JPG decode errors appear present in the server console either. |
|
(0012475)
James_Beejay (reporter)
2009-07-16 01:40
edited on: 2009-07-16 01:44
|
I have a 64bit server suffering the same issue.....
Debian 5
Mono 1.9.1 & 2.4
Both OpenSim.exe and OpenSim.32BitLaunch.exe have the same issue.
Works fine under 9841, but upgrade to 10035 causes the texture issue messages and a random crash (usually after many texture issues messages)
12:35:28 - [J2KDecoderModule]: JPEG2000 texture decoding failed for 91c9fe4f-6a6b-476f-808d-07d56577fe87. Is this a texture? is it J2K?
12:35:28 - Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 135 End: 0
Layer 1: Start: 225 End: 0
Layer 2: Start: 240 End: 0
Layer 3: Start: 255 End: 0
Layer 4: Start: 270 End: 0
12:35:28 - [J2KDecoderModule]: JPEG2000 texture decoding failed for 2ca94327-ad54-416e-8655-919625782ded. Is this a texture? is it J2K?
12:35:28 - Inconsistent packet data in JPEG2000 stream:
Layer 0: Start: 135 End: 0
Layer 1: Start: 194 End: 0
Layer 2: Start: 209 End: 0
Layer 3: Start: 224 End: 0
Layer 4: Start: 239 End: 0
12:35:28 - [J2KDecoderModule]: JPEG2000 texture decoding failed for 2c1ecd0d-8fa2-48e3-af8b-75fc24a723ac. Is this a texture? is it J2K?
|
|
(0012707)
lmmz (reporter)
2009-08-02 10:28
|
I still see texture errors in r10114 on my 64 bit region, also. COnsole says:
10:23:53 - Last packet end in JPEG2000 stream extends beyond the end of the file. filesize=276565 layerend=1685617016
10:23:53 - [J2KDecoderModule]: JPEG2000 texture decoding failed for 10c2ec04-d002-41b0-9c90-7747d3581546. Is this a texture? is it J2K? |
|
(0012708)
FrankNichols (reporter)
2009-08-02 12:10
|
I also see the texture errors, and new uploaded textures sometimes stall at the 64x64 resolution stage and never (as in hours) finish loading. |
|
(0012710)
jhurliman (manager)
2009-08-02 13:43
|
"I also see the texture errors, and new uploaded textures sometimes stall at the 64x64 resolution stage and never (as in hours) finish loading."
As a side note, there is no such thing as the "64x64 resolution stage". Textures in SL don't stream by resolution, they stream by signal-to-noise ratio improvements. LRCP progression is used instead of RLCP. It's such a commonly held myth that even many Lindens will tell you textures stream by resolution, but it is not true. |
|
(0012711)
FrankNichols (reporter)
2009-08-02 15:08
|
Hmm, on the 64x64 - I am seeing the texture in the texture console - it is showing 64x64 as the resolution of the texture - if and when the texture finally finishes loading it changes to 512x512 which it is actually. |
|
(0012719)
Bud Parnall2 (reporter)
2009-08-02 20:07
|
I have installed trunk with the same computer and different hard drive and loaded Ubuntu 9.04 32 bit rather than 64 bit. Mono 2.5. Did a restore of the database of the sim having trouble last night. Textures loaded perfectly and quickly. The problem is clearly with the 64 bit Ubuntu. I had previously had horrid texture issues. |
|
(0012722)
Kenjay (reporter)
2009-08-03 07:20
|
I'm seeing it with Fedora 9, 64-bit. The odd thing is, it didn't always do this, it started at a particular upgrade of the OpenSim code. If I back off to the old level, everything returns to normal without any database changes.
All I can tell from the backups is that the problem seems to have appeared between 9617 and 9859.
I'd appreciate any suggestions for getting it working properly again, other than going to a 32-bit OS. I use this machine for other things, and that's not feasible. |
|
(0012723)
Kenjay (reporter)
2009-08-03 07:27
|
Yes, just confirmed, level 9617 (restored from backup) works fine. I still get the decode error messages, but my hair and eyes are correct. On 9859 and newer they're both white. |
|
(0012838)
BlueWall (reporter)
2009-08-11 09:37
|
Does anyone have an idea where this is hanging? It seems that enabling "Atmospheric Shaders" is having some effect on on the textures too. Some of them (especially terrain textures) sharpen when it is enabled, or blur when it is disabled.
running openSuSe 11.1 64 bits here |
|
(0013092)
Bud Parnall2 (reporter)
2009-08-29 06:05
|
As of trunk last night the 64 bit vs 32 bit texture issues seem to have resolved. |
|
(0013178)
reacher (reporter)
2009-09-05 21:31
|
this still doesn't seem to be fixed. i just built opensim from trunk (as of 10 minutes ago), and am having the partially rezzed texture issue. this is debian x86_64 with Mono 2.4.2.3.
textures from the opensim library as well as newly uploaded textures remain "Loading..." forever, only partially rezzed. |
|
(0013179)
melanie (administrator)
2009-09-05 21:32
|
Sometimes, deleting the content of the j2kDecodeCache can help. Also, when using Flotsam, delete the region's cache. |
|
(0013181)
reacher (reporter)
2009-09-05 21:41
|
not using Flotsam, and clearing the j2kDecodeCache didn't help. but now i get additional errors in my console window in addition to the never fully rezzed textures:
21:39:30 - Last packet end in JPEG2000 stream extends beyond the end of the file. filesize=148 layerend=458752
21:39:30 - [J2KDecoderModule]: JPEG2000 texture decoding failed for 3d706261-3281-4409-81c4-36e655412a86. Is this a texture? is it J2K? |
|
(0013604)
BlueWall (reporter)
2009-09-22 14:19
|
Attached file, TextureData.sql contains some assets which tossed errors in the j2kdecoding with ...
16:45:34 - Last packet end in JPEG2000 stream extends beyond the end of the file
. filesize=4996 layerend=2097196
16:45:34 - [J2KDecoderModule]: JPEG2000 texture decoding failed for b8edc488-5db
0-4c2b-9c2c-11f9ae5c0eb7. Is this a texture? is it J2K? |
|
(0013623)
metaphorz (reporter)
2009-09-25 12:32
edited on: 2009-09-25 12:35
|
Same issue. The same opensim directory as used on a 32 bit Ubuntu
system running git f20a052 was copied to a 64 bit Ubuntu system.
SQlite is the db so all 4 dbs were copied intact. On the 32 bit
system, there are none of the yellow-highlighted errors of the sort:
[J2KdecoderModule] Last packet end in JPEG2000 stream extends beyond the end of the file, filesize = 11339 layerend=1344471584
[J2KdecoderModule] JPEG2000 texture decoding failed for <UUID>. Is this a texture? Is it J2K?
For the 64 bit system, there are numerous J2K decoder errors. Also, the
texture decoding does not appear to be texture-specific. On some occasions,
the same texture will not resolve (be fuzzy) and then resolve after restarting he server.
Tried flushing the client and j2k caches -- the [J2KdecoderModule] errors remain,
but it seems isolated to only the 64 bit installation.
Also, tried installing the bleeding edge head 2995d87, but same yellow
j2k errors..
|
|
(0013760)
metaphorz (reporter)
2009-10-05 20:41
edited on: 2009-10-05 20:44
|
In a version of opensim just downloaded (10/5/09), it would
appear that the CSJ2K library has been embedded into this version.
Here are some quick observations:
System: 2 x Quad-Xeon (64 bit Ubuntu 9.04 Server Edition)
1) For just about every texture downloaded, there is the thrown exception
of this sort (in yellow font):
23:16:01 - [J2KDecoderModule]: CSJ2K threw an exception decoding texture f70e335a-8985-d327-241b-4b52a482696d: Requested bitrate is too small.
23:16:01 - [J2KDecoderModule]: Failed to decode layer data for texture f70e335a-8985-d327-241b-4b52a482696d, guessing sane defaults
2) Texture decoding seems to be more efficient and faster with CSJ2K, and
many textures do resolve quickly, despite (1) above
3) There are still textures that never fully resolve, and remain fuzzy
The temporary workaround we are using currently is to have
each user overwrite their SL viewer cache with textures that
are known to be full resolution...this makes for quick
loading of textures and ensures that no fuzzy textures will
occur.
|
|
(0013827)
darjeeling (reporter)
2009-10-11 08:22
edited on: 2009-10-11 09:18
|
Experiencing this as a realy realy big problem as well, since last trunk update i've cleared all the caches, including local viewer cache.
Some textures rezz fine but most are totaly screwed.
Can anyone tell me about the CSJ2K "fix" ? When was it introduced in the trunk? Coz i've got the latest 0.6.7-post-fixes git version, and still seeing
[J2KDecoderModule]: JPEG2000 texture decoding failed for d3213adf-0e55-4895-abfe-f5e361fc3d6f. Is this a texture? is it J2K?
Instead of the CSJ2K messages.
And has anyone some clarification on the Texture Console?
What does it exactly mean when the State says SIM or DSK or NET or INI
I assume INI is initiate, NET is network (download) SIM i dunno, DSK, write to disk (as cache)?
|
|
(0014926)
minos (reporter)
2010-02-01 14:52
|
Same issue, some textures remain fuzzy fuzzy with CSJ2K saying "bitrate too small"
Issue presents on textures regardless of their upload format (PNG/JPG) and their resolution (1024x1024, 1024x512) |
|
(0014932)
miir01 (reporter)
2010-02-03 23:44
edited on: 2010-02-03 23:45
|
I'm experiencing the same issue. Some textures are not completely rezzed while others are. I get these log entries:
2010-02-03 20:42:09,698 WARN - OpenSim.Region.CoreModules.Agent.TextureSender.J2KDecoderModule [J2KDecoderModule]: CSJ2K threw an exception decoding texture e4dae6
91-3eec-c118-75e5-fa052f38dacc: Requested bitrate is too small.
2010-02-03 20:42:09,700 WARN - OpenSim.Region.CoreModules.Agent.TextureSender.J2KDecoderModule [J2KDecoderModule]: Failed to decode layer data for texture e4dae691
-3eec-c118-75e5-fa052f38dacc, guessing sane defaults
The strange thing here is that if I edit Appearance, then the textures are fine and seen at their full resolution. After leaving Edit Appearance they get blurry again. Same with "Rebake Textures". Textures are fine for a short time and then get blurry again.
I'm running this version on OpenSUSE 11.1 64bit:
Version: OpenSim 0.6.8 (Post_Fixes) (interface version 6)
I see that this issue is unassigned at the moment, is there not gonna be a solution? This is a very annoying problem and keeps users away from the grid.
|
|
(0014980)
Sylvia Sonoda (reporter)
2010-02-12 17:42
|
Some of the textures of my buildings in my sims are only loading half (blurry).
These same textures are no longer in my inventory (after the large inventory crash and repair a while back I think).
This maybe is the same issue?
Clear cash or another viewer does not make a difference.
It are always the same textures that stay blurry.
My sims run on a windows xp pro machine. I do not receive messages about it in the console window, I think. |
|
(0015041)
gjbivin (reporter)
2010-02-20 16:39
|
Just started noticing the problem with avatar clothing: running 0.6.8 Server on Windows Vista. Upgraded to the latest Emerald Viewer yesterday: 1.2.3.5 (1585), before which I hadn't seen the problem; but also now happens with Hippo 0.5.1 (LL 1.22.11). Also noticed that Rebake shows the texture properly for about 1 second, before going blurry again.
This is a problem for me because I'm using OS to make images for a graphic novel (this involves uploading hundreds of animations and textures, so being able to save all those L$ is a major reason for using OS). |
|
(0015145)
popi (reporter)
2010-03-11 17:27
|
can we says it's a 64 related problem?
Same on my ubuntu 64 bits.
2010-01-04 17:59:45,480 WARN - OpenSim.Region.CoreModules.Agent.TextureSender.J2KDecoderModule [J2KDecoderModule]: CSJ2K threw an exception decoding texture 6e0f6a1c-b1d6-4bed-a2b4-c10c9aa06c67: Requested bitrate is too small.
2010-01-04 17:59:45,480 WARN - OpenSim.Region.CoreModules.Agent.TextureSender.J2KDecoderModule [J2KDecoderModule]: Failed to decode layer data for texture 6e0f6a1c-b1d6-4bed-a2b4-c10c9aa06c67, guessing sane defaults |
|
(0015387)
Jet Dingo (reporter)
2010-05-01 06:50
edited on: 2010-05-01 06:54
|
I'll concur with gjbivin on this one.
I'm experiencing the same problems running OpenSim 0.6.8 on a 32 Bit System
Textures uploaded are not appearing 'fully loaded' with notifications when selected in the edit object window as "Loading" (in Emerald Viewer).
When applied to prim object or object attached to avatar, texture is blurry and appears to be not fully loaded.
Fault testing attempted so far:
Attempted texture export using custom viewer once imported into the simulation, export will not complete viewer side as texture is not recognized as fully loaded.
-----------------------------------------------------------------------------
Server System Specs Are:
Windows 7 (Ultimate) (32 Bit)
Intel P4 (3.00GHz)
2GB Ram
Simulator Environment:
OpenSim 0.6.8
Standalone Mode
Access via local LAN network only.
|
|
(0015388)
melanie (administrator)
2010-05-01 06:57
|
0.6.8 is ancient. Please consider upgrading. The texture decoding mechanisms have been rewritten twice since 0.6.8.
This issue is no longer relevant. Marking as fixed. |
|
TextureData.sql [^] (521,601 bytes) 2009-09-22 14:17 
