Cleaning up s3-data volumes

openBalena
1

Hi there,

Is there any cleanup/housekeeping script for the s3-data volume? In my current environment i am deploying around 6 releases a day each with 5-7 containers.

It seems like deleting old release on the openbalena-admin webui and api just deletes the reference instead of deleting the release too. And i am sitting right now at around 200gb for the open-balena_s3-data/_data volume.

Any help would be appreciated

2

I also would like to know how to do this. My server is currently 98% full. I need to delete old data to create room for new builds and releases.

I ran the following script to remove old releases. While I got a success message, the used space was still the same.

#!/bin/bash

# --- CONFIGURATION ---
APP_ID="<Fleet ID>"
AUTH_TOKEN="<Token>"
API_BASE_URL="https://api.balena.mydomain.com"
# ---------------------

# 1. Generate the list
# 2. Pipe into a loop
# tail -n +3 so that the last release is not removed 
balena release list alpha-autoFlow | awk '{print $2}' | tail -n +3 | while read -r COMMIT_HASH; do

    # Skip empty lines if any
    if [ -z "$COMMIT_HASH" ]; then
        continue
    fi

    echo "Attempting to delete release: $COMMIT_HASH"

    # Run the delete command
    # Note: We escape the $filter variable with a backslash so bash doesn't think it's a shell variable
    curl --location --request DELETE "${API_BASE_URL}/v6/release?\$filter=belongs_to__application%20eq%20${APP_ID}%20and%20commit%20eq%20'${COMMIT_HASH}'" \
    --header "Authorization: Bearer ${AUTH_TOKEN}"

    echo -e "\n-----------------------------------"
done
3

i have tried with that, but that curl request seems to only delete the references to the releases. And not the actual file. I had some success with a script i made, but today i found out it also deleted part of an active release and kinda made a mess of a working device.

So i am still here waiting to see if anyone on the balena crew could help out

4

I found this post. Not really something I want to try though.

5

As we are still testing to see if it will suit our needs, I went ahead into the s3 container and deleted all repositories and blobs to free up space. I figured I can just rebuild and redeploy. Worst case scenario, they are all R&D devices. Also deleted the releases using the script above (dunno if that actually does anything).

But this is huge handicap. This kinda shows that it isn’t ready for production (at least for our use case).