[zoppo]

i started a coop rendering on 3 machines:

1 - server and manager
2 - server
3 - server

nr2 crashed completely but nr1 showed no reaction at all.
nr2 is still in the render farm list.

job queue shows:
1 - waiting for others
2 - rendering (!)
3 - receiving mxi

and that's it ... no mxi has been written ... all lost

[tom]

Make sure,
- You installed same versions on each node.
- There are no firewall restrictions.
- There are available TEMP harddisk space on each node.
- There's no other application (like resident antivirus tasks) blocking access to coop files.

Also, do you encounter same problem after you close and then restart all the servers and the manager?

[zoppo]

Make sure,
- You installed same versions on each node.
- There are no firewall restrictions.
- There are available TEMP harddisk space on each node.
- There's no other application (like resident antivirus tasks) blocking access to coop files.

Also, do you encounter same problem after you close and then restart all the servers and the manager?

- fresh installed 1.5
- i don't think so. mxcl has full access and server rights granted. and they are all in the render farm window.
- lots of ...
- i got one av running (on the main pc nr1). but i don't think it blocks anything.

about restart: naturally i restarted pc2 that had crashed. but nr1 did not react at all - network status in mw stayed the same. so i had to restart nr1's manager and server. still no luck, as the manager only found the server running on the same machine.

so i restarted server on all three pcs and then started manager on nr2. they have been rendering now for 4 hours

but the important question is: what is supposed to happen when one of the clients crashes while rendering? and what happens when i decide to stop one of them while rendering in coop?

[deadalvs]

if You stop just one slave, it will wait for new work, and the job is progressing.

where did You create the .mxs ? studio or host app?

be VERY sure that every single texture has a path in unc form and also the output paths are correct in this form:

//computername/folder/texture.pic

don't use mapped drive names like T:\ !

[zoppo]

if You stop just one slave, it will wait for new work, and the job is progressing.

i created it in studio, but it was a resaved 1.1 scene.
every single texture was there as i used a pack&go directory.

BUT ... i used it as a mapped drive ...

AND ... while typing i stopped my coop render to test this - and it worked, the mxi was written on the mapped drive

BUT 2 - for a test i stopped one of the servers in the manager and then closed it - and its contribution was lost i think. is this behaviour wanted??

[deadalvs]

is this behaviour wanted??

* * *

hmm. i think i did also tests with mapped drives. the manager found them, the pc that the textures acually are on, but the third computer didn't start.

i also found my computers computing «something» but producing nothing. that was strange...