cancel
Showing results for 
Search instead for 
Did you mean: 

How to handle out-of-memory problems during repository load?

Former Member
0 Kudos

Hi all,

During load tests, we've loaded a lot of test data into a repository. The load itself worked, but as we unloaded the repository and loaded it again, we got an "out of memory" exception.

Now, as the test data wasn't valuable, we were able to just delete the repository. But what would be the options for valuable data?

Sure, a productive system should have enough memory installed. But imagine a defective client application that doesn't stop loading records into the MDM system.

I currently see two options - which are both not optimal in my opinion:

1. Transfer the repository to a system with more memory.

2. Access the database directly and extract the non-bogus data.

Are there further possibilities?

Best regards,

Dennis

Accepted Solutions (0)

Answers (1)

Answers (1)

Former Member
0 Kudos

Hey Denis,

I also had faced similar kind of problems. But do check the error report that flashes whenever the repository is unable to load.

It says that reduce the no. of records per fetch.

For me this thing worked coz there wasnt any need to move the repository to another system nor it was any need to access the database directly and extract non bogus data.

Check in the Server folder on the MDM server machine and open the <b>MDS.ini</b> file.

In this file search for the repository name which is giving problem and for that specific repository in the No. of records per fetch which would be by default initialised to 100 change that number to less than 100 like 80 or 85 or 50 and then save and close the MDS.ini file.

Now the MDS.ini file again need to be reinitialised so restart the MDM Server, so the ini file gets initialised.

Try dis out and I hope this solves ur problem

Regards Tejas ..................

Former Member
0 Kudos

Hi Tejas,

Thanks for your reply. In my case, the error message doesn't recommend reducing the numer of records to fetch.

I didn't have much time to investigate the loading issue (and it takes quite some time before the server runs out of memory...), but the interim results indicate that this parameter has only a minor impact in my situation, unfortunately.

I also found some other useful tips in the reference guides (such as setting disabling the "keyword" option for each field), but in the end, I doubt this will do the trick in extreme cases...

Best reagrds,

Dennis