Out of memory issue when saving network
24 views
Skip to first unread message
G. Papad.
Jun 26, 2023, 6:30:47 PM6/26/23
to visone-users
Dear all,
I'm loading a .csv file of about 160k links and importing on top of it one with node attributes from about 26k nodes.
When I try to save the resulting network as .graphml, I get an out of memory error message. The same holds even when I reduce the number of links to 40k.
Do you have any suggestions/ideas how to go about it?
Cheers,
Georgios
Müller Julian
Jul 2, 2023, 2:58:09 PM7/2/23
to visone...@googlegroups.com
Dear Georgios,
The number of nodes and links does not seem so high that you should observe this kind of message on saving the data.
It’s possible the Java runtime you are running visone on applies a very low memory limit. You can set a higher limit by starting visone from the terminal. If you are in the directory containing the "visone-2.26.jar" file (and the Java runtime is on the terminal’s PATH), then you could try to execute the following command in your terminal:
java -Xmx2048m -jar visone-2.26.jar
This sets the maximum amount of memory to 2GB. (You could also try to increase the limit even further by passing a larger number in the -Xmx argument.)
If that doesn’t help, then this problem is dependent on some feature of your data, but it’s hard to identify the specific cause without seeing it. If I had to make a guess, though, then I think it’s likely caused by some attributes you are loading. You can tell visone to not load some attributes by selecting “disabled” in the “status” row in the data preview section of the CSV import dialog. Or you could also try to delete some attributes after loading the CSV data in the attribute manager.
If neither removing some attributes nor increasing the amount of available memory solves the problem, then I need a minimal data sample that exposes the problem for debugging visone. You could send such a minimal example to my e-mail address: julian....@gess.ethz.ch .
Best,
Julian
The number of nodes and links does not seem so high that you should observe this kind of message on saving the data.
It’s possible the Java runtime you are running visone on applies a very low memory limit. You can set a higher limit by starting visone from the terminal. If you are in the directory containing the "visone-2.26.jar" file (and the Java runtime is on the terminal’s PATH), then you could try to execute the following command in your terminal:
java -Xmx2048m -jar visone-2.26.jar
This sets the maximum amount of memory to 2GB. (You could also try to increase the limit even further by passing a larger number in the -Xmx argument.)
If that doesn’t help, then this problem is dependent on some feature of your data, but it’s hard to identify the specific cause without seeing it. If I had to make a guess, though, then I think it’s likely caused by some attributes you are loading. You can tell visone to not load some attributes by selecting “disabled” in the “status” row in the data preview section of the CSV import dialog. Or you could also try to delete some attributes after loading the CSV data in the attribute manager.
If neither removing some attributes nor increasing the amount of available memory solves the problem, then I need a minimal data sample that exposes the problem for debugging visone. You could send such a minimal example to my e-mail address: julian....@gess.ethz.ch .
Best,
Julian
G. Papad.
Jul 3, 2023, 12:33:48 PM7/3/23
to visone-users
Dear Julian,
Thank you for suggesting the a solution. Initially I thought that the number of 160,000 links (and 26,000 nodes) might be the ones causing the problem. I will try the solution you suggest (also use visone on another machine) and see whether the problem persists. I will keep you posted on the outcome.
Thanks again!
Georgios
Reply all
Reply to author
Forward
0 new messages