Deploy Pipeline failes due to "no space left on device"

OBDDimal / variability.dev-database

Web service for sharing feature model instances and collaborative benchmarking.

7 stars 3 forks source link

Deploy Pipeline failes due to "no space left on device" #361

Open h3ssto opened 1 year ago

h3ssto commented 1 year ago

Log tag: https://github.com/OBDDimal/variability.dev/actions/runs/5303218282/jobs/9598683805#step:6:10

Steps:

[x] Decide on new maintainer for CI/CD
[x] Gain access to BWcloud-instance
[x] Make space :sweat_smile:
[ ] Add action of deploy to remove outdated images

Add command above of: https://github.com/OBDDimal/variability.dev/blob/9806375521da30088fca6b6290f969e2fc621b6f/.github/workflows/deploy.yml#L18

:heavy_check_mark: @Eixix Do you know what is the cause, how it can be prevented in the future?

Eixix commented 1 year ago

Will look into this tomorrow @h3ssto

h3ssto commented 1 year ago

Thank you very much! We will designate a new CI/CD master in our weekly tomorrow.

Eixix commented 1 year ago

The problem is having only 12 GB of space available on the device. The space is enough in theory, however everytime a pipeline builds a new Dockerfile, the old already built images will stay on the server. A temporary workaround for this is sshing into the server and executing docker system prune -a to remove any cached or old images/containers. I did this just now.

A longterm solution would be to automatically delete old images everytime a new one is built.

h3ssto commented 1 year ago

I guess it makes sense to only keep 2-3 versions of the image, thanks for your help @Eixix!