MRP runs with no jobs created for orders

Epicor ERP 10
1

On 10.2.200.10, and in .7 previously, we have been having MRP throw errors while running. These errors have resulted in no unfirm jobs creating for anywhere from 5-35 different orders during the run. If we push the orders to jobs, they always work fine, and have no issues.

The error logs do not show any error details, just say to ‘Look at the inner exception’. This is with a full regen MRP with 6/3 processors. I was able to run an MRP with no blank jobs this weekend by running the MRP process with 1/6 processors. To me this means we are getting SQL lock issues, and from my testing, this seems to be the case.

I did see some messages from earlier versions stating there were some patches available, but those were listed for earlier versions.

Does anyone have the same issues, and if so, is there a fix out yet? We have an open ticket with support on this, but have been trying to get a fix for about a month now.

2

Paul - How are you ‘seeing’ the missing jobs? I can test ours with a full regen in our 10.2.200.10 test environment…

If there were dead locks, you can turn on SQL Profiler to log/graph it.

3

We have a BAQ that shows us any orders that have no jobs to match them to.

I guess that is our next step is to look at SQL profiler, but it isn’t much help if it is an internal code issue with the way Epicor runs their MRP.

4

Does the MRP log tell you anything ?

5

The MRP log only tells us to look at the inner exceptions. I am almost 99% certain it is a SQL deadlock issue, since the problem goes away when we use 1 processor for generating the jobs. So the deadlock issue is an internal Epicor code issue in how they pull data, so running the profiler won’t help much, since I cannot change their queries.

6

Paul,
Can you provide some context to where in the log the error appears? Also is this basic or full MRP logs? Perhaps you can post the section of the logs that give this error?
Also, do you have ‘Recycle Jobs’ enabled?
How many parts do you have to process through MRP?
How long is MRP normally taking?
If it’s not more than an hour, I would try to reduce the number of processors to see if that helps. It could be a way to address any SQL deadlocks.
How would you gauge your SQL performance overall? This could be a symptom of your performance slipping and needing some maintenance.
SQL deadlocks during MRP would require a pretty specific data situation, such as missing data or a corrupt record or index.

I’ve seen MRP ‘see inner exceptions’ when there is an orphaned record that cannot be deleted for some reason. Somehow I was able to ID the Job and I tried to delete it manually which resulted in the same error and eventually I found that using the Actions > Scheduling > Remove From Schedule allowed the Job to then be deleted. Never found out what was wrong with the scheduling records.

1 Like