Permissions - access to writing file

jean-louis

Hi there,

We've just upgraded from RR 6.02.09 to 6.02.31 and I'm running into a problem.

Renders seem to work fine but are unable to write out frames.
I checked the RR/sub/log/c_... file and here's the pertinent line:

schoenberger · Site Admin Joined: 02 Mar 2005 Posts: 3786

Have you changed anything with the deamon? Was the client reinstalled?
Otherwise the execution user before and after the update would not have changed.

/TEST_1 is created by the rrSubmitter.
If you delete it is re-created by the rrServer.

The client logs the effective user and group it is running with at startup into the application log RR/sub/log/c_...
Please login with the same user and try to copy a file from the local machine to the network. Is that possible?
_________________
Holger Schönberger
Binary Alchemy - digital materialization

jean-louis · Posted: Fri Mar 14, 2014 5:51 pm Post subject:

Hi Holger,

I just ran the updater (6.02.31__updateFrom_6.02.00+), and then ran lx__rrWorkstation_installer.sh on each renderbox and rebooted them all (20 machines)

As far as user is concerned, the client log shows:

jean-louis · Posted: Fri Mar 14, 2014 6:22 pm Post subject:

Ok so I tried copying a file.

Firstly I couldn't log in (rruser is created by lx__rrWorkstation_installer and for linux there is no password) so I logged in as root and removed any need for a password for rruser (passwd -d rruser), then logged in.
(BTW Is this behaviour correct? i.e. no password and thus unable to login?)

Anyway - I had no problem copying a file, so that's not the issue.

Any ideas?

Cheers,
Jean-Louis

jean-louis · Posted: Fri Mar 14, 2014 6:26 pm Post subject:

For good measure, and to see if you can find any clues, here is the rrControl log:

jean-louis · Posted: Fri Mar 14, 2014 7:12 pm Post subject:

More info...

Not sure what I might have done in the meantime (I really haven't been fiddling with any config files) but suddenly I'm getting this happening whenever I try rendering:

schoenberger · Site Admin Joined: 02 Mar 2005 Posts: 3786

The global.sh files are executed before the renderer starts.
But the renderer writes the files to the local drive, the client tries to write the fileserver.
So render logs and setenv will not help.
The error is not printed by the renderer, it is printed by the client.

Please check the application log file RR/sub/log/C_... for a client start before you reinstalled the deamon via the rrWorkstationInstaller.
Is there a different user/group?

You can also zip and upload your RR/sub/log via www.RoyalRender.de/upload_r.php then I can take a look.

>rruser is created by lx__rrWorkstation_installer
lx__rrWorkstation_installer does not create any user.
It uses something like
su <yourusername> or sudo <yourusername>
_________________
Holger Schönberger
Binary Alchemy - digital materialization

jean-louis · Posted: Fri Mar 14, 2014 8:01 pm Post subject:

Unfortunately there are no logfiles with info from before except for windows machines (where the user is System both before and after the update) and they're working fine anyway.

In any case I've uploaded the sub/log folder.

so you know:
Rdroid1 - 20 are the renderfarm nodes and are all Linux
DG1 - 12 are workstations and all windows7 64bit

Thanks,

Jean-Louis

jean-louis · Posted: Fri Mar 14, 2014 8:17 pm Post subject:

schoenberger · Site Admin Joined: 02 Mar 2005 Posts: 3786

No logfiles from before is not that good. So we cannot check if the user was different before the deamon reinstallation.
_________________
Holger Schönberger
Binary Alchemy - digital materialization

schoenberger · Site Admin Joined: 02 Mar 2005 Posts: 3786

Do you have some old jobs from before the update?
Those render logs show the user.

The folder /RR/sub/stats/ is new and it seems that one has wrong permissions.
The file/RR/sub/cfg_global/clients.ini does not have any issue. Do you see any rights difference if you compare them from a linux machine?

Ok, the issue can have two reasons:
-Wrong user after the installation
-RR creates the folder by a different application. E.g. it was created by rrServer, now it is created by rrSubmitter.
_________________
Holger Schönberger
Binary Alchemy - digital materialization

jean-louis · Posted: Fri Mar 14, 2014 9:13 pm Post subject:

schoenberger · Site Admin Joined: 02 Mar 2005 Posts: 3786

http://###########chat
_________________
Holger Schönberger
Binary Alchemy - digital materialization

jean-louis · Posted: Fri Mar 14, 2014 9:46 pm Post subject:

Hey Holger - a big thanks by the way!

jean-louis · Posted: Tue Mar 18, 2014 2:43 pm Post subject:

Hi Holger,

Thanks for all your help on Friday.
I'm afraid I have to revive the thread since my problems are persisting...

I'm not getting the problem with permissions anymore, thanks to your intervention, so that has definitely been resolved.

The issue now is that whenever there is more than one machine assigned to a job there's a constant amount of crashes, and gradually machines are being disabled from the job.
Frames *are* being rendered, but obviously it takes a lot longer.
However if there is only one machine assigned, there is absolutely no issue at all.

Note that this only happens with Arnold renders. Mental ray renders work fine.
I have re-installed Arnold (SItoA 3.01) but it doesn't make any difference.
I have also looked around if there's any .lock files being created in the workgroup but there's none.
This happens both with linux and windows machines.

Is this something you have seen before?

By the way here is a snippet from the log:

schoenberger · Site Admin Joined: 02 Mar 2005 Posts: 3786

Is local render is enabled for the job?
And if you render AoV, are they written to the same main output exr or directly to a different fileserver path?

If the issue is somewhere in the Arnold render, then I do not know why it does not print any message.
There is no rrErr: error message somewhere else in the log file? (as standard and error log can get out of sync).

Does it happen with all kind of scenes? e.g. a scene with a lot of reflective/refractive spheres and high AA (render times high enough to get a comparison, but loading fast and not using any external files like textures)?
_________________
Holger Schönberger
Binary Alchemy - digital materialization

jean-louis · Posted: Tue Mar 18, 2014 3:22 pm Post subject:

Hi

Local Render Out is enabled in the submitter, yes
Just an rgba .exr, no AoVs

Here's some more errors/warnings although most are timer info (I've not enclosed as code so that I could bold the error lines):

///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
S 19| rrFilereplace 7.0.b2+6.2 Copyright(c) Holger Schoenberger. All rights reserved.
S 20|
S 21| ----------------------------------------------------------------
S 22| Error: input file does not exist! '/usr/local/RR_localdata/renderer_prefs/softimage/lx_x64/2012.0\Data\Preferences\default.xsipref'
S 23|
R 0| \"/RR/bin/lx64/rrResetexitcode\"
R 1| +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
R 2| ++++++++++++++++++++++++++++++++++++++++++ START ++++++++++++++++++++++++++++++++++++++++++++++
R 3| ++++++++++++++++++++++++++++++++++++++++ 7.0.b4+6.2 ++++++++++++++++++++++++++++++++++++++++++
R 4| +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
R 5| \"/usr/Softimage/Softimage_2012/Application/bin/XSIBATCH.bin\" -processing -thread 8 -script \"/RR/render_apps/scripts/kso_softimage.py\" -lang Python -main rrStart -args -argAll \" PyModPath: /RR/render_apps/scripts, SName: /PROJECTS/TEST_ARNOLD_RR/Scenes/TESTB_6_JL.scn, Db: /PROJECTS/TEST_ARNOLD_RR, Pass: Default_Pass, SkipFrame: false , FName: /usr/local/RR_localdata/renderout/A/TEST_ARNOLD_RR/Render_Pictures/Default_Pass/TESTB_6_JL/<Layer>. , FExt: .exr , FrStart: 40, FrEnd: 57, FrStep: 1 , FrOffset: 0 , Verbose: 0, \"
R 6| =======================================================
R 7| rrErr: Scan for Custom SPDLs: Elapsed time (ms): 75.71
R 8|
R 9| rrErr: Loading User plug-ins: Elapsed time (ms): 0.34
R 10| rrErr: Loading User addon plug-ins: Elapsed time (ms): 0.12
R 11|
R 12| License information: using [Processing]
R 13| rrErr: Loading Workgroup plug-ins: Elapsed time (ms): 1044.00
R 14|
R 15| ' ERROR : <InspectObj>, This command can't be used in batch mode. - [line 3633 in /usr/Softimage/Softimage_2012/Application/DSScripts/uixsiscripts.vbs]
R 16| ' ERROR : - [line 1262 in /usr/Softimage/Softimage_2012/Application/DSScripts/uixsiscripts.vbs]
R 17| AddProp "Custom_parameter_list", "Scene_Root", siDefaultPropagation
R 18| Command failed, returned -2147418113
R 19| ' ERROR : - [line 62 in /XSI/WORKGROUP64/Addons/agBend/Application/Plugins/agBabel.js]
R 20| ' WARNING : PyQt not found, etRenamerPyQt will not be loaded.
R 21| rrErr: Loading Workgroup addon plug-ins: Elapsed time (ms): 1446.00
R 22|
R 23| ' INFO : 4337 - Plug-in name conflict resolved:
R 24| ' Resolution mode: Versioning
R 25| ' Loaded plug-in:
R 26| ' <name>
R 27| ' <version>
R 28| ' <path>
R 29| ' Conflicting plug-in(s):
R 30| ' <version> <path>
R 31| '
R 32| ' INFO : 4337 - Plug-in name conflict resolved:
R 33| ' Resolution mode: Versioning
R 34| ' Loaded plug-in:
R 35| ' <name>
R 36| ' <version>
R 37| ' <path>
R 38| ' Conflicting plug-in(s):
R 39| ' <version> <path>
R 40| '
R 41| rrErr: Loading Factory plug-ins: Elapsed time (ms): 1445.14
R 42|
R 43| rrErr: Loading Factory addon plug-ins: Elapsed time (ms): 1320.55
R 44| rrErr: Loading setup plug-ins: Elapsed time (ms): 0.03
R 45|
R 46| ' INFO : 4337 - Plug-in name conflict resolved:
R 47| ' Resolution mode: Versioning
R 48| ' Loaded plug-in:
R 49| ' <name>
R 50| ' <version>
R 51| ' <path>
R 52| ' Conflicting plug-in(s):
R 53| ' <version> <path>
R 54| '
R 55| ' INFO : 4337 - Plug-in name conflict resolved:
R 56| ' Resolution mode: Versioning
R 57| ' Loaded plug-in:
R 58| ' <name>
R 59| ' <version>
R 60| ' <path>
R 61| ' Conflicting plug-in(s):
R 62| ' <version> <path>
R 63| '
R 64| ' WARNING : 3000 - Multiple compounds have the same name: fgr_neck_setup_contrain (1.0). Using: /XSI/WORKGROUP64/Addons/Biped_Generator/Data/Compounds/OLD/Character_Setup/FGR_Neck_Setup_Constrain.xsicompound. Ignoring: /XSI/WORKGROUP64/Addons/Biped_Generator/Data/Compounds/Character_Setup/FGR_Neck_Setup_Contrain.xsicompound
R 65| InstallCustomPreferences "ArnoldRenderPreferences", "Arnold Render"
R 66| rrErr: XSI Init: Elapsed time (s): 18.34
R 67|
/////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////

Yes it seems to happen with all scenes.

Jean-Louis

schoenberger · Site Admin Joined: 02 Mar 2005 Posts: 3786

No, these errors do not have anything to do with the crashes.
Sorry, you have to ask in the SitoA list. You have updated SitoA as well before the issues started, right?

So far I do not see something.
The scene file is loaded once, you can try to enable local scene cache then nothing is loaded from the fileserver. If you do not use textures.
SitoA should load all files at startup as well, but I am not sure.
_________________
Holger Schönberger
Binary Alchemy - digital materialization

jean-louis · Posted: Tue Mar 18, 2014 3:46 pm Post subject:

OK thanks Holger - I'll keep investigating.

Cheers,

Jean-Louis