This issue has been popping up on a few slaves here and there. Nothing has really changed in what we’re doing and I can’t see any issues with the actual machines. It’s kind of weird, but not a show stopper at all.
Error occurred while writing report log: Could not find a part of the path ‘\renderfarm\DeadlineRepository6\reports\jobs\92\6\5307c72127f3da0c7498d926.bz2’. (System.IO.DirectoryNotFoundException)
Hello Jay,
So this looks like a possible rogue slave issue. Can you verify whether “\renderfarm\DeadlineRepository6\reports\jobs\92\6” exists? If not, how far into the folder’s path do you get? That should be a pretty clear indication of the issue. Thanks.
So this is happening today again. I looked for the directory and it stops at \renderfarm\DeadlineRepository6\reports\jobs. There’s actually nothing in the jobs directory. What’s even weirder is that this totally happens on random slaves.
So it seems like one particular frame is the one that is having issues. No matter what slave it gets onto it gives the same error. Is there a way to purge this issue? Or maybe I need to resubmit it as a new job?
Hello Jay,
Can I have you verify if there is a slaves folder in the Repository root? If there is no filed beyond “\renderfarm\DeadlineRepository6\reports\jobs” then there is definitely an issue, because all of your job reports should be in there. If you have a slaves folder in your repository root, then the hostname of the folder(s) in there will identify machines that have a 5.x slave on it, which is wiping out your reports folder. If not, we need to figure out what is wiping that out. It is very odd that only one job is having issues, though, as based on the design of 6.x, if you have nothing beyond that path, you wouldn’t be able to read or write and job reports.
I’m out of the office today, but I can log in remotely, so my responses may be spotty. It’s interesting, there is a slaves directory in the repository root, but there is only one directory in there and it’s from an OSX slave that’s been giving me a ton of problems. This slave will start sending me stalled emails every few hours or so even though the slave was never launched. The only way to stop it is to start the slave then kill it. I’m not sure if this is related but it’s curious.
James
It sounds like this OSX slave is still running Deadline v5.x? Which would explain the issues you have been seeing! Can you uninstall Deadline client software from this machine or just turn it off to confirm this is the machine which is causing you this issue?
I’ll have the artist check his machine. What exactly should I be seeing in those directories?
You shouldn’t have a “slaves” directory in Deadline 6. If you do, then whatever slave named directory is in this directory is still running Deadline v5, which has a side-effect of purging your job reports. Hence the error.
Aaaaaahhhhh, now it all makes sense. OK, I’m having the artist look into it now. So I can safely delete that slaves directory?
Yes, but there’s not much point as it’s going to keep coming back at you until you sort out the machine causing the issue. 
OK that’s what I wanted to know.
My artists says that Deadline 6 is the only version on his Mac. I’ll double check it in the morning. Is there any way of detecting something like this?
Nevermind, he has a PC and a Mac at his desk. He looked at the wrong one. 
I am having a similar issue, I saw that one of the nodes was in the slaves directory I went to that node and removed Deadline 5 and deleted the slaves folder from the repository. But I am still having the issue are there any other steps that i could take?
If you delete this slaves directory and wait a while, does it get re-created? If so, same slave name? Different slave name? Maybe you have more than one node still with Deadline v5 installed on it? If not, then this issue is resolved. Deadline v6 repository installer as a MAJOR version, must be installed to a fresh location compared to Deadline v5 repository installer. All MINOR releases, can be installed over the top of a pre-existing repository. When you installed Deadline v6.0 or v6.1 (as a fresh install) did you install it to a fresh location, ie: NOT the pre-existing Deadline v5 directory?
HI,
The folder is not being recreated. I am restarting the node to see if that helps.
you mean the repository? repository was a fresh install to a new machine.
Correct. I think you just need to get some restarts in here and you should be good. In production, I always liked to get a restart of all the farm in at least once a week. It normally helped all manner of things. I configured the Power Management to do the reboots for me automatically. (It would restart on current task completion, if the slave/machine was busy rendering when the time came for a weekly reboot). Strange how you installed a fresh to a different machine and yet, the old v5 slave could still find the new location? Maybe you shared the new repository path out as exactly the same path? Maybe as a DFS redirect / global namespace? Anyhow, your reports directory should start behaving now.
Yeah I have no idea how that happened. But I didn’t install it on that machine, my IT department did…so who knows. As for the power management stuff, it’s something I’m looking into but at the moment our IP situation is a bit off. There are some slaves that I can’t access or send commands to remotely. But that’s a good idea to have a restart scheduled each week.
This issue is popping up again this weekend. It seems that one machine keeps throwing the same error over and over and eventually it knocks the job off the farm. I already removed Deadline 5 from the problem machine and there is no more slave directory in the repository. Yet the issue persists.
I also ended up reinstalling the repository. I am not sure if that will ruin your job history or anything like that . the error has been gone though since I did that.