AWS Thinkbox Discussion Forums

Error:FailOnExistingMaxProcess - ???

Hi, we’re almost finished setting up our 10 node Deadline farm and for the most part things are running but I still have a lot of jobs that fail prematurely, for no clear reason.

Please see this screenshot:

On several nodes I’m getting an error saying: Error: FailOnExistingMaxProcess is enabled, and a process 3dsmax with pid 6272 exists - shut down this copy of 3dsmax to enable network rendering on this machine
at Deadline.Plugins.PluginWrapper.StartJob(String& outMessage, AbortLevel& abortLevel)

…but I did a remote connect on these machines and there is no other copy of 3DS Max running so I don’t get the reason for the error.

Also, I tried resubmitting these jobs with the Override on Existing Max Process set to true, then false and I get the same error either way. This is really a fairly simple job where I am trying to save out partitions from a pflow to a prt sequence.

Does anyone have any tips on how to either prevent these errors or get Deadline to ignore them? Thanks!

If you are saving partitions, are you running concurrent tasks on the same machine? That would launch multiple instances of 3ds Max, and could explain the issue. If the Concurrent Tasks option is disabled in the Krakatoa Deadline submission panel, then we would have to dig deeper. (It could happen with running multiple Slaves on the same machine, but that requires manually starting them, so you would know about it, right?)

When the Override is set to True, the second value should be set to False, meaning “DO NOT FAIL on existing 3dsMax process”. In the screenshot, both are set to True.

Hi Bobo,

Thanks for responding so quickly. I pasted a screen grab below showing the Deadline job properties and the submission settings. I have concurrent tasks set to 1:

We only have one version of Slave running per node. I just remote connected to some of the nodes to confirm this.

Thanks for the clarification for the setting for the Fail on Existing Process dropdowns. I resubmitted the job with those settings corrected and I’ll check back in a bit to let you know if the job failed or was successful. I’ll keep you posted…

The main location to change the Fail On Existing Process setting (instead of overriding it) would be as SuperUser mode under the Configure Plugins > 3dsMax in the Tools menu of the Deadline Monitor. It actually defaults to False, and it exists mainly because in the early 2000s, most computers had 2 CPUs and launching 3ds Max on a workstation where 3ds Max was already running in UI mode was a bad idea. :wink: So we had this turned on in early versions of Deadline to prevent the Slave from rendering Max jobs if Max was already opened on an artist’s machine…

I don’t see a reason why you would want it to be set to True nowadays.

Hey Bobo, thanks for the explanation. I went into Super User mode and changed that setting so hopefully I won’t run into that issue again.

I am still getting some other errors and I’m not sure what might be causing these:

So I’m confused by these errors:
Error: RenderTask: Unexpected exception (Failed to open the max renderer.
and
ERR: [08824] [09468] Licensed number of users already reached. Feature: krakatoa-render License path: @flexlm-svg1; FlexNet Licensing error:-4,132 For further information, refer to the FlexNet Licensing documentation, available at “www.flexerasoftware.com”.

This was on a simple 30 frame test render and it did complete the render OK despite the errors. However, I’m getting the same errors on my partition job, and while it’s running OK for now I hope I don’t leave today just to come back tomorrow morning and see that it failed after only reaching 3% completion (which is what’s been happening).

As far as the Licensed number of users already reached error goes my initial thought was that we had an old license still sitting on our license server, and perhaps that was getting picked up instead of our current license. We just upgraded our Thinkbox license to include 10 Deadline render nodes and we previously only had 2 render nodes running with Deadline’s free version. So I thought that maybe the license server was giving out the old and incorrect license. I checked with our IT guys and they confirmed that they removed the old license file so I’m not sure why we’d still get this error. I should also mention that while I got this error on the simple 30 frame test render that job did successfully render on 6 of our 10 render nodes (2 are offline), so that makes me even more confused about what the story is with the error.

The errors says the “krakatoa-render” license count limit is reached. This is not about the Deadline licenses, but the Krakatoa Render licenses. However, I see that you have 10 of them, too, so I am not sure why you are getting a license error. It is possible that some of the nodes are simply slow in returning the license - if a render node drops off and some other joins in, it is possible that an 11th node would try to get a license before it is available.
Are you rendering on exactly 10 render nodes?

Hey Bobo, thanks for your response. I’m trying to resolve an issue that I’m guessing is causing the Krakatoa license error. Please see the attached screen grabs:

Now that we have ten Deadline and Krakatoa licenses we are running them on the slaves deadline1-10. DOVA is my workstation and it is the PC that I submit jobs from. Tapir is my coworker Tyler’s workstation and he submits jobs from there. When we only had a Krakatoa license with 2 render nodes each we were each using a free version of Deadline with the two nodes, and at that time tapir and dova were used as render slaves, but now we are not trying to do that. I tried deleting the slaves DOVA and tapir in the slave list at the bottom of Deadline monitor, but they are still showing up in the whitelist/blacklist list in the Machine Limit panel of the Job Properties.

As I said, I tried deleting tapir and DOVA as slaves and they don’t show up in the slave list in the console.

I tried running a repository repair and at first it said it loaded 10 slaves, then it said it loaded 11.

I tried changing my workstation’s startup options to prevent Deadline Slave 10 from launching at startup. I then tried restarting and Slave started anyways, and then it showed back up in the slave list in Monitor.

Is there any way I can have the Deadline Client run on my workstation so I can submit jobs but not use my workstation as a Slave? I tried to see if I could uninstall slave but it looks like that’s part of the deadline client, which is what I need to submit jobs. What am I missing in regards to deleting these unused, unlicensed slaves from the repository? Even if these did show up, if I had them blacklisted then why would I get license errors if I’m not trying to use more than 10 slaves?

You do not have to disable the launch of the Deadline Launcher in the Windows Startup, just right-click the Deadline Launcher icon and make sure the “Launch Slave At Startup” is unchecked. Also, it does not matter if your workstations are listed as active, idle or disabled Slaves in the Monitor, since you have already whitelisted the job to only render on the 10 render nodes. You could also create a “Krakatoa Partition” Pool and add the 10 machines to it, and submit your Krakatoa Partitioning jobs to that Pool.

The blacklist entries in the right column will stay there until you move them to the left, and if they don’t exist anymore, they will disappear. But they play no role if the respective Slaves are not listed in the Slaves panel anymore, so don’t worry.

In general, you don’t really have to tweak your Deadline settings anymore, they appear to be right. What is misbehaving is the FlexNet license management, and it is possible that the error message is in fact bogus.

I will ask around and get back to you a bit later after I get a better understanding on the exact steps Krakatoa performs to acquire a license, and return it. I wonder if the returning of a license gets delayed and the SAME Slave attempts to get a license for the next frame again before the license server has updated the return.

Where is the license server installed? Is it a dedicated machine, one of the workstations, or somewhere else?

Thanks for the help Bobo. Yes, please keep me posted on what you find out regarding the Krakatoa licensing.

Just to confirm, I didn’t see any check boxes in the Deadline Launcher for Launch Slave at Startup, but that option and Restart Slave if it Stalls were highlighted in green. I clicked on them and the green highlight went away, I’m assuming that means those options are now disabled.

Our IT guy that was helping to set this up installed the Repository on the first slave on our network: Deadline1 …Would that location for the repository cause any issues?

Thanks again!!

I asked about the License Server, but running the Repository OR the License Server on a render node is a big NO-NO. Imagine you are rendering in V-Ray and there is 0% CPU left, how is the Repository or License Server going to serve requests from the other Slaves?

We generally recommend having a dedicated machine just for the Repository, and possibly a separate one for the License Server, but with 10 nodes you can probably share the machine. In fact, you could probably use one of the Workstations for the Repository and the License Server, and let the Render Nodes just render and do nothing else. Assuming the Workstation is not used to render on all CPUs too often… :wink:

As for the Launch Slave checkbox, yes, those icons are check items, and unchecking them disables the features.

Thanks Bobo, I’m sending you an email about this.

Privacy | Site terms | Cookie preferences