Greetings,
We have Archivematica v1.8.1 running and we're trying to perform the following scenario:
1. Create a transfer of the folder "record-1" which looks like:
# record-1/my-data.txt
Some test data for my example...
# record-1/metadata/metadata.csv
filename,dc.title,dc.description,dc.identifier
objects/,My record title,My record description,record:1
2. The transfer goes through ingestion and stores an AIP. If we download the AIP, the metadata is correctly included in the METS.xml file (attached as "record-1-d746d26f-cad3-4838-9d22-6ef3610efb57-initial.7z"):
<mets:dmdSec ID="dmdSec_2">
<mets:mdWrap MDTYPE="DC">
<mets:xmlData>
cterms.xsd">
<dc:title>My record title</dc:title>
<dc:description>My record description</dc:description>
<dc:identifier>record:1</dc:identifier>
</dcterms:dublincore>
</mets:xmlData>
</mets:mdWrap>
</mets:dmdSec>
3. We now want to update the metadata of the AIP, so in the "Archival packages" page we issue a "Re-ingest" with "Metadata only" (as described
in the docs)
4. In the Ingest tab we click on the "page and pencil" icon to add the new metadata files
The metadata.csv has been changed in the transfer source folder as follows:
# record-1/metadata/metadata.csv
filename,dc.title,dc.description,dc.identifier
objects/,My new record title,My new record description,record:1
5. We proceed with the ingest by selecting "Do not normalize" (as described in the docs), and everything finishes normally
6. We download the AIP again from the "Archival Storage" page (attached as "record-1-d746d26f-cad3-4838-9d22-6ef3610efb57-reingest.7z"), but the METS.xml has not been updated with the values from the new metadata.csv. We can see the metadata.csv file being included, but now the directory structure has been changed, with an extra level of nesting being added (highlighted file, is the new "metadata.csv"):

Basically what we want is to be able to update the METS.xml metadata of the AIP possibly multiple times after the original transfer and creation. If that is not possible, we would like to have an AIP which doesn't grow in depth every time we add some metadata.
Also, to draw out the bigger picture: the initial AIP creation and sub-sequent metadata updates are basically part of our automatic workflow that "syncs" records from our "live" repository to our Archivematica instance for preservation.
Thank you in advance for your help