Age | Commit message (Collapse) | Author |
|
use threading for interprocess communication. Don't stall slave communication when rendering/baking process output blocks. This enables running slow baking and rendering jobs correctly without the slave disconnecting from the master. It also makes slaves much more responsive to cancelling jobs on the master.
add save on job option (default false) to save the current file before sending a rendering blender job.
|
|
|
|
typo prevented returning proper content-typeé
Thanks to Philippe Van Hecke for bug finding.
|
|
- distributed point cache baking
- (baking fluids or anything else needs to be added but is not a hard task)
- master support getting all results as a zip (rendering, baking or otherwise), available as a link on the job's page in the web interface
- framework to support multiple result files per frame/task (needed for baking)
|
|
|
|
wip feature: distributed point cache baking. Distributed baking works but results are sent back to master/client yet. Feature is disabled in the UI for this reason.
new feature: job and slave tags, enables filtering slaves for specific jobs
jobs are dispatched to a slave only it has no tags or all the job's tags
Render jobs have the tag "render" by default while baking jobs the tag "baking"
UI: Web interface additions to reflect tags and job subtypes (render/baking)
bug fix: reseting a completed job correctly resets the status to queued
|
|
- Support for dynamic painting point cache (patch by Philippe Van Hecke)
- Fix job log name generation to not exceed filename limits when using large chunks
- Cleaning up duplicated code for point caches, fluid and multires between client and repath module
|
|
correcting smoke domain modifier type restriction
Patch by Philippe Van Hecke
|
|
- fix handling of pointcache dependencies on slave (correct repathing) and on client (bad frame dependency range)
- add Force Upload option on master to force dependencies upload and download (don't reuse local copies outside of slave cache)
- add links back to main page on job web page
- client limit dependencies to rendered frame range (it used to add all point cache frames)
Thanks to Philippe Van Hecke for raising the issue by email.
|
|
[#29648] patch for netrender: correctly send smoke cache domain
Both by Philippe Van Hecke
Render engine support was expended to be able to specify non-builtin engines.
|
|
|
|
edits), removed so this wont happen again.
|
|
fix error when sending thumbnail to master
|
|
Fix for #26867
Hunted down and debugged with great help from Geoff Murphy
|
|
* Scan for master failed and gave an script error.
|
|
|
|
|
|
Based on patch [#28108] by Wintch Analyzer
|
|
Reported by Tobias Kummer. Thanks! :)
|
|
by Wintch Analyzer
moving master working path initialization before connection creation
also added error checks and prints to debug an invalid internal state (related to another tracker entry)
|
|
exception safeguard in render engine
fix module import broken by someone else
|
|
- remove/comment unused variables
- remove unused imports
- fixed some bugs using incorrect variables
|
|
fix problem on web interface with blacklisted slaves no longer in slave list
|
|
better stream handling when uploading files
Might solve the problem with OS X masters
|
|
Reported by Ramon Nieto
Added a better error message when the current file is not saved.
|
|
Add rna callback to server address property so that it correctly checks if the server exists when the value changes.
|
|
Report by Wintch Analyzer
Refill the rna blacklist list every refresh (safer that way).
|
|
|
|
|
|
the right fix at all.
|
|
|
|
|