Hi,
maybe I am just overlooking something, but I could not find a way to set the length of the output buffer for the slave log. Sometimes when a render job is quite long and thus outputs a lot of progress messages, the information from the beginning of the render log is not visible anymore. Is that maybe restricted to the single last log file?
For there is no other way to access the log of a running render job I really need a longer log output.
Thanks
Felix
The whole log can be found on the machine that’s currently rendering. There are actually two files you can check out:
The first is the Slave log itself, and you can find it using this page in the docs:
docs.thinkboxsoftware.com/produc … html#slave
The Slave log is going to interleave all of the output which will be difficult if your job is set to use concurrent tasks. The second log is going to be best as it’s written by the render thread. It’s named as “deadlineslave_renderthread_[thread id]-[slave name]-0000.log” where “thread id” is the index for that concurrent task, and “slave name” is the name of the Slave that was writing it.
Those will always be present as it’s our output for the job reports and it will get decorated with machine information and compressed over to the Repository. Occasionally the Slave may be forcibly closed which will leave those files around. They’ll be overwritten when the next job comes around.
Now, why do we only have 5,000 lines of scrollback? Mainly it’s memory reasons. We thought that if that wasn’t enough, it was unlikely that 10,000 would be much better. I’m not aware if we secretly support bumping the number up, but I’ll look into it. I personally doubt we do.
Ok, I know about the slave logs and I understand all the technical aspects and reasons.
But, and especially from our operators/artists view, it is mandatory to access the full render log of a running job/task from the monitor interface.
At this time this limit is hard coded, and it would take awhile to get that code changed and tested to allow for more, however as Edwin pointed out, even doubling this would likely not be of much help. Is it possible for the users to just remote to the slave and pull up the slave logs?
Hm, too bad. First we are using a mixed Linux/Windows environment. At the moment I try to get some windows remote access up and running but that is not that easy as I thought. On Linux it is a bit more simple, but more technical (ssh/command line only).
Apart from that is is not really a practical sollution to teach all the artists how to technically remote login onto two different kinds of systems and search for some specific files manually. Furthermore the logfiles on windows are located in the c:\Program Data… folder which is not even visible to normal users by default.
Sorry, I don’t want to be offensive but this is something mandatory and other rendermanagers support it.
Which render managers support live-streaming the log from the beginning? Mostly a curiosity thing on my part.
We can look into maybe opening the file read/write and having the Slave stream it from disk instead of holding it in memory.