Possible Memory Leaks during importing

Hello everyone!

I’m importing a large amounts of DICOM images (500GB) into dedicated Testing Orthanc server with Postgres indexing.
I’m using python script found in your documentation.

Orthanc is contaneraized with Docker (osimis/orthanc:20.9.5 image).

What I’m noting is that used RAM is increasing, so this is a problem especially for large ingest of data.

After 45min of script running and tens of GB sent, this is the situation:
NAME | CPU | MEM | PIDS
orthanc-service 27.39% 5.147GiB / 31.37GiB 16.41% 64

orthanc-db 0.03% 5.59MiB / 31.37GiB 0.02% 8

Any solutions?

Thank you

Jacopo

UPDATE:
After an hour RAM usage is stable at 6GB.

Now the questions are:
why does orthanc need so much memory?
could this behaviour affect other operations (with multiple tasks like zip generating and downloads)?

Thank you so much

Jacopo

Hello,

I think that the following FAQ in the Orthanc Book should answer your question:
https://book.orthanc-server.com/faq/scalability.html#controlling-memory-usage

Summarizing, you should upgrade “osimis/orthanc:20.9.5” to a version above 20.12.2. As of writing, the latest version is “osimis/orthanc:21.2.0”.

HTH,
Sébastien-

Thank you so much!
I’m upgrading my orthanc service to newer version.

However, I noted that used RAM was stable 8GB, also after completing the importing process with old 20.9.5 (Maybe useful for you).

Thank you again

Jacopo