0006752: [WARNING]: EOC marker not found. Codestream is corrupted.

(0024324)
hack13 (reporter)
2013-09-09 08:05

I can normally get this issue to go away by manually clearing out my script's cache folder. Try that and get back to me.

(0024325)
aiaustin (developer)
2013-09-09 09:11
edited on: 2013-09-09 09:13

I deleted the ScriptEngines folder which I do preserve between upgrades and restarted without it. I cannot see the [WARNING]: EOC marker not found. Codestream is corrupted.

But oddly, I then put back the original ScriptEngines folder and on restart I again cannot see the [WARNING]: EOC marker not found message! Would you expect that?

I upgraded from r/23570 to the latest r/23613 and ran with the ORIGINAL Script Engine cache folder and the message showed again on first run after upgrade.

(0024326)
hack13 (reporter)
2013-09-09 13:19

I can't really say anything, other than perhaps maybe there is a hidden file or something that gets dropped in during the initial upgrade from one git to the next in that folder that upsets it.

(0024333)
hack13 (reporter)
2013-09-12 00:50

Upon further investigation, this appear to happen when bullet physics are turned on for me, when using ODE I do not get this error (as often, just relates to scripts then)

(0024334)
aiaustin (developer)
2013-09-12 01:58

Interesting. I confirm that all my regions run BulletSim.

Now, this may not be related, but Robert Adams (BulletSim developer) did once tell me about it being important not to have a carriage return on the end of the config-include/GridCommon.ini file and (unusually for OpenSim config files) the GridCommon.ini.example does end right after the text on the last line.

I always thought this was odd and likely to lead to fragile behaviour as most people would add an extra line on the end of any configuration file.

It may be unrelated, but could it be that some end of script or asset processing for BulletSim is different to using ODE.

(0024336)
kenvc (reporter)
2013-09-12 08:54
edited on: 2013-09-12 09:08

I was seeing this a lot when using Bulletsim. I switched back to ODE today because Bulletsim was using a lot more memory than it was last week, but I have still been seeing the error occasionally that is reported in this mantis.

I most often see it very shortly after a message about Warp3D not being able to decode a texture and sometimes the instance freezes when this happens.

(0024348)
hack13 (reporter)
2013-09-15 01:32

I am seeing this more often, I am going to do more investigations.

(0025029)
butcharnold (reporter)
2014-01-20 13:00

I see this error only when running Bullet Physics and Warp3d Map module.
We are using 7.6PF.
I can get rid of the error by changing either one... it seems the error only happens for us when we have both enabled.

(0025030)
Richardus Raymaker (reporter)
2014-01-20 13:02

I see it sometimes after loading OAR file. but it disapeared spontanic to

(0031517)
aiaustin (developer)
2017-01-01 04:00

A note that this issue still seems to be present in 0.9.1.0 dev master at 1-1-2017. It would be good to try to get a bit more diagnostics at the point where this message is put out to the OpenSim.exe console (but its not logged in any at to openSim.log).

(0031742)
Ferd Frederix (reporter)
2017-04-08 23:39

Try changing mySql.ini to have a much larger max_allowed_packet.

Large vars can chew up enormous amounts of space for the land texture.

# required for large files and land textures
max_allowed_packet = 1000M

(0031743)
aiaustin (developer)
2017-04-09 01:58

Interesting. Thanks for the suggestion Ferd. I don't actually have var regions on my own grids, but its worth a try of there is no serious downside of giving a very high max_packet_allowed value.

max_allowed_packet for my grids at the moment is 16MB, and I had a note from ages ago that 5the largest asset (at the time) on OSGrid was 45MB

Notes
(0024324) hack13 (reporter) 2013-09-09 08:05	I can normally get this issue to go away by manually clearing out my script's cache folder. Try that and get back to me.

(0024325) aiaustin (developer) 2013-09-09 09:11 edited on: 2013-09-09 09:13	I deleted the ScriptEngines folder which I do preserve between upgrades and restarted without it. I cannot see the [WARNING]: EOC marker not found. Codestream is corrupted. But oddly, I then put back the original ScriptEngines folder and on restart I again cannot see the [WARNING]: EOC marker not found message! Would you expect that? I upgraded from r/23570 to the latest r/23613 and ran with the ORIGINAL Script Engine cache folder and the message showed again on first run after upgrade.

(0024326) hack13 (reporter) 2013-09-09 13:19	I can't really say anything, other than perhaps maybe there is a hidden file or something that gets dropped in during the initial upgrade from one git to the next in that folder that upsets it.

(0024333) hack13 (reporter) 2013-09-12 00:50	Upon further investigation, this appear to happen when bullet physics are turned on for me, when using ODE I do not get this error (as often, just relates to scripts then)

(0024334) aiaustin (developer) 2013-09-12 01:58	Interesting. I confirm that all my regions run BulletSim. Now, this may not be related, but Robert Adams (BulletSim developer) did once tell me about it being important not to have a carriage return on the end of the config-include/GridCommon.ini file and (unusually for OpenSim config files) the GridCommon.ini.example does end right after the text on the last line. I always thought this was odd and likely to lead to fragile behaviour as most people would add an extra line on the end of any configuration file. It may be unrelated, but could it be that some end of script or asset processing for BulletSim is different to using ODE.

(0024336) kenvc (reporter) 2013-09-12 08:54 edited on: 2013-09-12 09:08	I was seeing this a lot when using Bulletsim. I switched back to ODE today because Bulletsim was using a lot more memory than it was last week, but I have still been seeing the error occasionally that is reported in this mantis. I most often see it very shortly after a message about Warp3D not being able to decode a texture and sometimes the instance freezes when this happens.

(0024348) hack13 (reporter) 2013-09-15 01:32	I am seeing this more often, I am going to do more investigations.

(0025029) butcharnold (reporter) 2014-01-20 13:00	I see this error only when running Bullet Physics and Warp3d Map module. We are using 7.6PF. I can get rid of the error by changing either one... it seems the error only happens for us when we have both enabled.

(0025030) Richardus Raymaker (reporter) 2014-01-20 13:02	I see it sometimes after loading OAR file. but it disapeared spontanic to

(0031517) aiaustin (developer) 2017-01-01 04:00	A note that this issue still seems to be present in 0.9.1.0 dev master at 1-1-2017. It would be good to try to get a bit more diagnostics at the point where this message is put out to the OpenSim.exe console (but its not logged in any at to openSim.log).

(0031742) Ferd Frederix (reporter) 2017-04-08 23:39	Try changing mySql.ini to have a much larger max_allowed_packet. Large vars can chew up enormous amounts of space for the land texture. # required for large files and land textures max_allowed_packet = 1000M

(0031743) aiaustin (developer) 2017-04-09 01:58	Interesting. Thanks for the suggestion Ferd. I don't actually have var regions on my own grids, but its worth a try of there is no serious downside of giving a very high max_packet_allowed value. max_allowed_packet for my grids at the moment is 16MB, and I had a note from ages ago that 5the largest asset (at the time) on OSGrid was 45MB

Issue History
Date Modified	Username	Field	Change
2013-08-30 05:42	aiaustin	New Issue
2013-08-30 05:46	aiaustin	Description Updated	View Revisions
2013-09-09 08:05	hack13	Note Added: 0024324
2013-09-09 09:11	aiaustin	Note Added: 0024325
2013-09-09 09:13	aiaustin	Note Edited: 0024325	View Revisions
2013-09-09 09:13	aiaustin	Note Edited: 0024325	View Revisions
2013-09-09 13:19	hack13	Note Added: 0024326
2013-09-12 00:50	hack13	Note Added: 0024333
2013-09-12 01:58	aiaustin	Note Added: 0024334
2013-09-12 08:54	kenvc	Note Added: 0024336
2013-09-12 09:08	kenvc	Note Edited: 0024336	View Revisions
2013-09-15 01:32	hack13	Note Added: 0024348
2014-01-20 13:00	butcharnold	Note Added: 0025029
2014-01-20 13:02	Richardus Raymaker	Note Added: 0025030
2017-01-01 04:00	aiaustin	Note Added: 0031517
2017-04-08 23:39	Ferd Frederix	Note Added: 0031742
2017-04-09 01:58	aiaustin	Note Added: 0031743

Relationships