Migrate using watch folder?

Questions, comments, discussions. Over time certain topics might be moved to their own category.
Post Reply
recklessop
Posts: 2
Joined: Fri Sep 13, 2019 2:45 pm

Migrate using watch folder?

Post by recklessop » Fri Sep 13, 2019 2:50 pm

Hello.

I want to migrate my old Mayan install (which doesn't use Postgres and is pretty old (2.7 maybe? I haven't logged into it for a while)

I have a new Mayan deployment using docker-compose, with Postgres, watchtower, etc.

I would like to move all of the documents from the old setup to the new. (this is just for personal documents so i dont have to have a fileing cabinet)

Instead of trying to export the database/convert to Postgres as well as an update to Mayan 3.2, could I just copy all of the items in the document storage folder and rename them with a .pdf ending, then drop them in a watch folder on the new installation?

seems like this would cause the new Mayan to OCR them and put them into the database etc.

Obviously, this will suck up CPU like crazy and probably take a while. but I don't have any metadata on my documents since its just for personal stuff anyhow.

Any downsides to doing it this way?

TIA!

recklessop
Posts: 2
Joined: Fri Sep 13, 2019 2:45 pm

Re: Migrate using watch folder?

Post by recklessop » Sun Sep 15, 2019 8:35 pm

As an update I was able to SFTP all the documents from the old server to the new server.

I then used the rename command to rename all of the objects to PDF.

I then created a watch folder from that directory and told Mayan to check for files.

It took it about 20 minutes to pull in all 1200ish items. However, inside of Mayan I now see 2513 documents. For each document imported I see two objects. Im not too worried about it because I have them all as a "Migrated" document type.

It has been about 24 hours since I imported them all, and tesseract and postgres are still chewing up two cores of the CPU. I have about 1 GB of free ram out of the 16 GB in the box (5.4GB is showing as buffer / cached).

Just as a note on the CPU this is only an Intel Celeron J1900 So I expect OCR to take a while.

Post Reply