I have to update 4500 employees standard hours and reading this from filestorage and having memory issues using the LM Transaction node. Anyone have any thoughts/ideas on updating 4500 employees without hitting the memory threshold?
are there any impact with the output ? if no you can ignore the error.
I could ignore the error; however, I know there have been discussions from Infor that they will be disabling any flows that have this memory issue. I'm not quite sure on the time line when this will take place; however, I would like to get this addressed.
I would need to see the flow but there are a variety of things that you could do.1. How are you reading, looping through the data? There are a variety of ways to do this and which path you choose impacts the efficiency of the flow.2. How are you updating the employees? What is the exact landmark transaction? Are you using a filter or populating the full keys of the set/index? 3. You could always grab the records, store in array with a parent flow, pop X keys off start a child workunit to run a smaller set and just run this in several workunits. I would target an execution time less than 10 minutes as that is how long flows are given in the event a ZDT patch needs to be applied anyway. Feel free to open a ticket with Infor Support for the memory alert and provide your .lpd file; we don't make changes for you but we can review the current design and make more specific recommendations on how to get below the alert level.
Hi Russel -
1. I'm looping through the data, calculating all hours for the year of all employees, dividing the hrs worked by number of weeks worked, then writing it out to a file.
2. On the second loop, i'm reading in the file created from the first loop using the LM transaction node to update and populating the keys, no filter.
3. I was thinking I'd probably have to break this flow up into two, one to do all the calculations and create files at certain intervals. Those files would consumed by file channel to kick off another flow to process the records in smaller chunks using LM transaction node as you stated.
I will open a ticket with Infor Support to review and I greatly appreciate your quick response and feedback.
Thanks Jose, please mention in the incident that you were discussing this in the Infor Community and Russ is requesting the incident to be assigned directly to himself.
I did mention that we were discussing on Infor Community when initially opening the ticket. I've updated the ticket to have it assigned to you.
