Migration to file index to db

jacopo.deluca · August 2, 2023, 10:31am

Hello everyone,

I’m trying to migrate a dockerized Orthanc instance from file indexing to postgres.

Reading documentation, better solution in my case is ImportDicomFiles.py which I’ve already used.

The idea is to create new dockerized instance with postgres configuration alongside the old one with same shared dicom storage and execute ImportDicomFiles.py.

I’m pretty sure in the past the migration/importing has generated same directory hierarchy, so I’ve been expected this behaviuor again with overwriting dcm file in dicom storage, letting me two instances of Orthanc with different indexes and sharing storage, with no disservices and little issues only with very latest studies (not crucial now).

Unfortunately during some recent tests I noted ImportDicomFiles.py executes a migration but generates different folders, so this means I cannot avoid disservices and mainly I’ll need double of the space besides the fact I’ll have to perform a delete of old studies which are not overwritten.

Is actually like this?
There is a way to perform what I need?

Hope I’ve been clear

Thank you so much

Jacopo

Enkhsaikhan · August 3, 2023, 12:59am

hello maybe you have to try this
this is fast and can skip files

gist.github.com

https://gist.github.com/jphdotam/21581fc4a205072ecf30d2c0c846f117

migrate_orthanc.py

# this is inspired by the official ImportDicomFiles.py - however, this is MUCH faster because
# it uses multiple workers to dramatically reduce IO bottlenecks
# it doesn't re-instantiate the Http() and headers for every file, rather every folder

# it does assume the input folders are structured in the orthanc format, e.g. /00/00/0000a8768bd86...
# and expects PATH_TO_SYNC to be something like:
## "/path/to/orthanc/root/*"  for all root folders
## "/path/to/orthanc/root/1*" for all folders beginning with 1
## "/path/to/orthanc/root/23" for root folder 23 only

This file has been truncated. show original

jacopo.deluca · August 7, 2023, 8:52am

Thank you so much, it is not exactly what I need but it is a good starting point to increase speed and also space-saving during importing.

Thanks again

J.