From: Jeff Jasper <jjasper@(email surpressed)>
Subject: Re: Rush 102.42a9d runs verified to operate normally on Lion (OSX10.7)
   Date: Thu, 19 Apr 2012 21:30:48 -0400
Msg# 2225
View Complete Thread (7 articles) | All Threads
Last Next
We are all good now! yeah!

On 4/19/12 5:35 PM, Greg Ercolano wrote:
On 04/19/12 17:18, Jeff Jasper wrote:
Just checked if I run the submit scripts from the 10.5.8 queue manager
workstation then everything fires off correctly.

	OK, so sounds like you solved the previous problems.

Also Lion workstations where the user account is hosted on the server
work fine, and local accounts that are on the box work fine.

What doesn't work strangely is a home account that is synchronized with
the server so the account is local on the box but sync back to the
server. We have a few people setup this way since some apps don't like
Lion server homes.

	Hmm, sounds like a different issue.

	If the job submits but doesn't run on the farm,
	hit 'Cpus' in irush for that job, to see if the cpus
	the job is requesting is having trouble running jobs
	as that user on the remote machines.

	A common error is to see in the NOTES field of the
	'Cpus' report is:

		501: unknown uid

	..where '501' might be any numeric value, usually
	representing the UID of the user who submitted the job.
	This error means that uid is not valid on the remote
	machines, so rush can't run the renders as that user.

	If you get that error, the fix is to make sure that user
	has an account on all the render nodes with that same UID.
	(It's important to have consistent uids for accounts)

	Some companies prefer to force all renders to run as
	a particular user, such as a 'render' user with some
	fixed uid/gid. (To do that, you'd set the forceuid/forcegid
	values in the rush.conf file)

	Otherwise, please try to follow up with error messages from
	the Cpus report, Frames report, or the rushd.log file, as
	it should be indicating the error in one of those locations.

Solved it on the two stubborn machines by doing a reinstall of rush on the boxes instead of just running the install script again. Working like a champ now


So we have a viable workaround. Only change I am seeing on all
workstations since the Lion upgrade is the Modo submit script is
throwing up an error at startup. You can run it fine still after
bypassing the error screen.

The error is:
input: /Network/Servers/"serverhostname"/"users home
directory"/.rush/.submit-modo-last: Line 1: unknown field '####ModoCommands'

	Hmm, sounds like some odd text is getting into
	the history file that keeps track of your last submission.

	Has the script been customized at all, or is there
	anything unusual in how the submit script is invoked?
	(ie. invoked by another script that tells the submit script
	what the default options should be?)


No it is just the standard setup, been afraid to mess with it since it is working. It does seem like something with the history because on subsequent launches there is no error, just the first time it is launched. I will shoot you the script in email. I am not seeing anything out of the ordinary. It is working great with Modo 601 too.

Thanks,
Jeff


Last Next