Memory and Dashboard issue with some patch Windows 7 PCs

[robert_pelle]

[ FlexSim 18.0.3 ]

I am on Windows 7 SP1 and IT threw some automatic Windows and .NET patches last week, and a model that we have been running with minor input changes started hanging, with issues with memory climbing to consume all system resources as well as issue with Dashboards. This was the same on another Windows 7 PC, but a similar PC seemed to work fine as did a PC running another on Windows 10. Issue cannot be traced to a unique patch (see attachment) and to remove variable of model, I made the basic tutorial model and added a couple of dashboards and they threw many exceptions and the model would consume all memory.

We are standardized on 18.0.3 long term version and I removed and reinstalled FlexSim but problem is still recurring. I am thinking of upgrading to 18.0.5.

Any reported issue with Windows or .NET patches reason to believe upgrading version will help? Thanks. windowspatchissue.xlsx

[JordanLJohnson]

Can you save the tutorial model that breaks and post it?

[philboboADSK]

"We are standardized on 18.0.3 long term version and I removed and reinstalled FlexSim but problem is still recurring. I am thinking of upgrading to 18.0.5."

The best way to find out would be to actually install 18.0.5 and find out. You can easily uninstall 18.0.5 and reinstall 18.0.3 if you want to go back.

Are you having trouble with using modules? Can you drag a conveyor into the model without an exception? Or add a ProcessFlow?

Windows 7 has an issue with loading the module dlls if you are missing the Windows security update KB2533623. See Error on using modules.

[robert_pelle]

Apologies if this is redundant, but tried to post reply and it asked to renew my login, so not sure if it sent or not.

Jordon, I am attaching a simple model based on tutorial#1 that has exceptions on startup. If I delete and re-add the dashboard, it consumes all available memory.flexsimtutorialexception.fsm

The systems were updated several times from Wednesday through Friday last week, so I have not yet been successful in replicating the problem from scratch.

Phil, I can add Process Flow without any errors, but have a graphics issue with the conveyors. I am guessing this is related to the graphics card, as the conveyors get added but cannot be seen, unless I uncheck the render setting in system properties. In other words, I can only see them if drawn without the rollers. Regarding the modules fix, I read the referenced post, which recommended 18.1. Is the same fix as 18.1 in 18.0.5 LTS?

Thanks

[philboboADSK]

Yes, if conveyors are invisible, you just need to update your graphics driver. See Conveyors are invisible in render mode.

The referenced modules post was talking about a change made in 16.1, not 18.1. If you can add a ProcessFlow without throwing exceptions, then you have the necessary Windows update. You can ignore that post.

[philboboADSK]

I was able to replicate the memory explosion and crash running that model with 18.0.5 and 18.1.1.

I don't think it has to do with your computer's windows updates.

We'll debug it and let you know what we find out.

[JordanLJohnson · Accepted solution]

In the model you have posted, the flowitem bin is missing. The crash is caused because the Source is trying to create an item, and it cannot determine which item to create.

Do you have any guesses on how the flowitem bin was destroyed? It should be present in every model. When you start a new model, is the flowitem bin still there?

[robert_pelle]

Phil, good to hear you have replicated a memory issue. Is that related to the FlowItem Bin issue that Jordan mentioned?

Jordan, That is indeed odd that the FlowItem Bin is missing and I haven't a clue how it happened. The FlowItem was there however on the original REAL model which was hanging in the first place and motivated me to experiment with the simple tutorial model. For this real model, I can now make it happen consistently with batch scripting on more than one system. I can however replicate the Dashboard exceptions in the tutorial model. If I take a perfectly good model, created with the Enterprise license, and try to run it in RunTime it says "Operation not allowed in this version", and corrupts the model, so that if I save it off it unrolls these exceptions the next time you open the model. This is quite unexpected as I routinely run thousands of models with batch scripting for optimization, using the runtime licensing.

exception: Exception caught in TreeNode::receivemessage()
exception: Exception Caught in ObjectFunction103__project_exec_commandlist_firecreateevent_node object: MAIN:/project/exec/commandlist/firecreateevent/node i: MAIN:/project/exec/commandlist/firecreateevent/node class: MAIN:/project/exec/commandlist/firecreateevent/node

exception: Exception caught in TreeNode::callMemberFunction() c++/dll execution. Throwing... MAIN:/project/exec/commandlist/firecreateevent/node c: MAIN:/project/exec/commandlist/firecreateevent/node i: MAIN:/project/exec/commandlist/firecreateevent/node thisClass: MAIN:/project/exec/commandlist/firecreateevent/node
exception: Exception caught in evaluation of MAIN:/project/exec/commandlist/firecreateevent/node
exception: Exception Caught in ObjectFunction361__project_library_statistics_ChartTemplate_behavior_eventfunctions_onPostSave object: /Tools/ChartTemplates/State Gantt class: MAIN:/project/library/statistics/ChartTemplate

exception: Exception caught in TreeNode::callMemberFunction() c++/dll execution. Throwing... MAIN:/project/library/statistics/ChartTemplate>behavior/eventfunctions/onPostSave c: /Tools/ChartTemplates/State Gantt thisClass: MAIN:/project/library/statistics/ChartTemplate
exception: FlexScript exception: VIEW:/nodefunctions/savemodelrestore

[philboboADSK]

"Phil, good to hear you have replicated a memory issue. Is that related to the FlowItem Bin issue that Jordan mentioned?"

Yes, I explained that we could replicate it, and Jordan explained the cause of the issue.

18.0.5 has the following release note:

• Fixed exceptions with opening chart templates while over the model limit.

I suspect that the cause of the exceptions and corruption you are seeing has been fixed. You should upgrade to 18.0.5.

[robert_pelle]

newtutorialwithdashboardin805.fsm

Thanks for the tip on 8.0.5, Phil - I ran the simple model and it does indeed open up cleanly and there are no exceptions. However, even if I remake the dashboard from scratch, in the Enterprise version, the dashboards do not display in the Runtime version. So, the errors are indeed gone, but there appears to be a remaining licensing issue with the new dashboards and Runtime licensing. That's a significant issue for us, as we build models in Enterprise and then many times run them under multi-core optimization using runtime licensing.

Having said this - this simple test was just my attempt to debug the original problem, which persists on 8.0.5, which is the runaway memory consumption of certain models, at least when run in batch. Since you folks said that a missing flowitem bin can cause that symptom, we will have another look at our code to see if perhaps somehow we are corrupting the model.

[robert_pelle]

just a follow-up. This post brought to light some issues with runtime licenses and dashboards, which I ran into while creating a simple model to help debug the issue that I had been seeing in regards to runaway memory usage on a complex model. Some of this is fixed with 8.0.5, but there are still problems with running dashboards with runtime licenses, even though there are no exceptions thrown. I hope this can be address in a future patch to the LTS. But in the ORIGINAL complex model that I was debugging, which motivated the study with the simple model, we finally tracked it down to a problem in our batch file which we were using to run FlexSim in an optimization loop. So, as it turned out, it was a rare occurrence and unrelated to system patches or the dashboard issues. Thanks.

[philboboADSK]

Thanks for the update. I added a case to the dev list about the chart runtime issue. We're looking into it. I'm glad you got your original issue worked out.