Linux slaves get stuck in "idle" state.

Deadline
1

Hi all, I have a problem. The slaves can only run their work submitted by theirselves, i.e. Machine A submit a maya scene a.mb, A could render it correctly, however other machine could not fetch the job and keep the state in idle!
I guess it’s some problem related with linux auth, so I chmod the jobs directory under DeadlineRepository to mod 777, but it still not work.

Both the repository and slaves use centOS, The version of deadline is 5.0. hope any one can helps…

2

Hi,
Please can you post your full job log report and/or any error reports / messages as well if anything appears on screen?
Mike

3

Hi Mike,
the job report is completely an idle report:(a longer version has been updated)

I have a further chmod, which chomod the entire DeadlineRepository to 777, then all the task state become rendering and render Slave is the idle slave, which is really weird, I mean, the slave is still idle, in both the slave output window or on the monitor, however all the rest task change into Rendering state.

4

And there’s not any error.
BTW Yesterday I install the repository to a Win plateform and get those form of error too, all the rest tasks’ state became Rendering and the linux slave kept idle. I changed the plateform to linux today, but have the problem of this thread :frowning:

5

Hello,

I wanted to ask if you can turn on the job candidate filter(the checkbox option in the upper left of the slave window) and click on the jobs to verify the monitor sees those other machines as able to do the job. Theoretically any job that can do the job should be able to pick it up. Is there any indication the other slaves are even trying to pick up the job and getting errors? Can you verify there is no use of pools, groups, or limits that would exclude the other machines? Thanks.

6

Yes I’m pretty sure all those machine can be seen from monitor. Pool/Groups/Limits have been left to the default setting of None, seems no mis-setting could happened at this stage. I setup deadline on Windows before with no problem, so I guess most the problem could probabily located in the install and configure stage.

Hope the problem could be solved today, I have a real deadline. :frowning:

7

Hey guys, finally I figure it out!

  • All the clients use one specified user account.
  • All the clients’ time set to identical. I have encounter this problem 3 years before, when their time’s difference is too large from each other, it became idle, strange…

Thanks for your time!

8

Well, it’s great to see you got it worked out, but oddly curious that it happened. I am certain, though, that this has been fixed in later versions.

Cheers,