Digdag was crashed due to h2 database full

[revanna7226]

I set a Digdag with H2 DB and ran for 10 months. Digdag worked fine upto 10 months for our project and crashed, from the available evidences we claimed that digdag is crashed due to database full. When we checked the H2 database tables and data we had created multiple projects in the application. Our intention was to check the diskspace required by database for 1 year. Hence we tried to delete the other projects records from the database. but h2 console is not querying actual data.

SELECT * FROM session_attempts WHERE project_id = 1 is resulting 0 records. But we can records in the table with Project ID is 1 when we view table.

What is happening here? Why this conditional query SELECT * FROM session_attempts WHERE project_id = 1 is not working?

How to check how much diskspace is consuming per session? If we know this we can calculate diskspace required for 1 year.

Thanks for reading... Revanna

[hiroyuki-sato]

Hello, @revanna7226

If digdag projects returns the project which id = 1, It contains project data in your database.

digdag projects
2021-09-06 22:40:30 +0900: Digdag v0.10.2
Projects
  name: hoge
  id: 1
  revision: 273de50f-0967-4838-8de6-5b80f631838b
  archive type: db
  project created at: 2021-09-06 22:40:10 +0900
  revision updated at: 2021-09-06 22:40:10 +0900

Use `digdag workflows <project-name>` to show details.

This ER and this comment may help. (But, I've never tried it. I recommend you to backup the database before try)

digdag_er.pdf

[hiroyuki-sato]

digdag: 0.10.2 h2: h2-1.4.192.jar

digdag.conf

database.type = h2
database.path = ./digdag.db

java -jar /tmp/h2-1.4.192.jar

• JDBC URL: jdbc:h2:/private/tmp/ddd/digdag.db/digdag
• User Name: empty
• Password: empty