That thread you posted Kayleigh is probably the correct issue here. In that link, the problem is coming from uwsgi, a component used in the Archivematica Storage Service.
Jesús Garcia Crespo (an Artefactual developer) first tried running the storage service with gunicorn and found that this particular problem went away.
It appears to be an issue that came with a newer version of uwsgi, the problem did not exist 6 months ago, to my knowledge at least. uwsgi has a pretty fast development cycle, we have not tracked down the exact version where this problem crept in. Switching to gunicorn solved the problem, and so at Artefactual we have focused on that as a solution.
The 1.5.0 release of Archivematica , when installed using the .deb
packages, still uses apache + mod_wsgi for the dashboard, and nginx +
uwsgi for the storage service.
We'd like to transition both apps to use nginx+gunicorn by default for the 1.6.0 release.
Hector Akamine, one of the sys admins at Artefactual, added support for gunicorn to the ansible role we use at Artefactual to deploy Archivematica. If you are using the deploy-pub repo (
https://github.com/artefactual/deploy-pub.git ) and you add a couple of variables to your vars_singlenode.yml , like so:
archivematica_src_ss_gunicorn: "true"
archivematica_src_am_dashboard_gunicorn: "true"
when doing a fresh install of Archivematica, you will end up with the Storage Service and the dashboard both deployed using nginx + gunicorn. This has resolved the downloading large aip problem for all the cases we have seen so far.
I'd be interested to hear from anyone that is trying this combination. If you have a current production Archivematica server, and you want to switch, you would have to do a bit of configuration work, but if you follow the examples in the pull requests where support for gunicorn was worked out in the ansible role
https://github.com/artefactual-labs/ansible-role-archivematica-src/pull/54it should be possible to switch any Archivematica system over to gunicorn.