We just migrated e-cienciaDatos from v4.19 to v5.10.1 two weeks ago. We needed one work day for the production server migration.
Our dataset files are in a disk server (not S3 or similar) out of the main Dataverse server.
- At first we migrated our Dataverse instance in the test server and put here our customization and translation.
- We have write a little migration guide at the same time. This guide is more specific for our particular case that the general guide and we detailed here the problems found.
- We created a war with our v5.10.1 customization.
- We advice to our librarians to not create datasets the day of the production server migration (only librarians can create datasets in our installation)
- The migration day, we connected the disk server to the test server to avoid stop the service for 6-8 hours (we profited to make several operating system updates). We changed the main e-cienciaDatos url to point to the test server. The only problem of this approach is that we loss the usage statistics of the time that the production server was out of service.
- Before the production server migration, we had all needed files organized for Dataverse version (payara versions, solr, tsv files, dataverse war files, ....)
- After the migration and customization, we connected the disk server to the production server, we had changed the e-cienciaDatos url to point the production server, and we had said to the librarians that they can create datasets.
Tell me if you want more information.
Regards,
Juan