Hello there, if you're able to start your instance you should be able to export all your data using as csv
And create a fresh latest new twenty instance where to import them
Please note that you're a very old twenty instance and I would not recommend trying to upgrade to the latest instance
As said the shortest path would be either exporting all your data using native export as csv feature ( either from api directly or UI ) or using pg dump and restore manually on a fresh new latest version twenty instance
#Old Twenty Data Recovery
1 messages · Page 1 of 1 (latest)
I imported CSV to a new instance and it broke the metadata or something like that because I can see that companies and people are populated with my data but twnety 1.12 cannot rebuild page layouts, list views, widgets, filters, tabs, UI Metadata!. So maybe the way to go is downgrade then upgrade again?
its seems newer versions wont rebuild metadata locally, only via cloud?
You need to configure the metadata the same way it was on your previous instance
Nothing behind a paywall between cloud and self host experience here
My previous instance just broke, I didnt know the version. Cant remember. It was a true frankenstein. But I was able to pgdump the untouched database. I could not make it work with any version fom 0.21 to 0.50. Failing queries in the login! So I exported people, companies, tasks, notes, opportunities to CSV and imported via SQL injection to a new instance of Twenty 1.12.0. That did broke my UI.
What should I do If I need to maintain data binding between people, companies, taskes, etc? Can I install fresh and just import the CSVs via front end? Should I downgrade and try to migrate the database?
Thanks in advance!
Should I downgrade and try to migrate the database?
Most complicated way IMO
If I were you I would try to transform my pg_dump to a csv format
You can restore it in a db and export rows using third party pg tools
Then I would re-configure my metadata to match the ones I have in pg_dump
To finally re-import the metadatas
Please note that between 0.20 and 1.12 there's more than a year of twenty upgrades
Pg tables and schemas evolved a lot
That's why I would make my data twenty agnostical by converting it to csv
Sorry to make this longer, What king of pg tools should I use? I did exported CSVs but the SQL inserts were the real issue, correct?
If the workspace and its users are new can I still use the original metadata?
TY!
No worries all good
What king of pg tools should I use?
If you already have your data as csv none
I did exported CSVs but the SQL inserts were the real issue, correct?
Do you mean on a latest twenty instance csv import ?
If the workspace and its users are new can I still use the original metadata?
You will have to configure the metadata ( data model settings ) from the UI manually on the new instance to be iso with the on you're about to import
Hey! Thanks for the last interactions
I did create new seld hosted instance. But before I imported all my csvs to Twenty CRM online, checked that all relations were kept, and them exported them again to ensure max relieability.
Then I installed a new self-hosted instance using the latest image,
I successfuly imported all the companies, more than 3k records.
Then I tried to import Persons and I got a 2025-12-10T19:19:18.933264597Z query failed: INSERT INTO "workspace_3ba7mp1nx7ljf89clgin9vraf"."person" AS "person" altogether with this long insert string.
How can I fix this?
In the front-end the message was something like: this record already exists
I was able to upload using the original CSV but not with the CSV exported by Twenty online. Can we call it a bug?