Closed nerok closed 7 months ago
Hello @nerok thanks for filing!
So I had a look with one of my colleagues from the Turbinia side and your theory is right, we have not updated that variable in the init-configmap.yaml
and that path doesn't exist within the API pod itself. We will put out a fix for this soon but in the meantime, a workaround can be to either execute into the API pod and create that directory or download the Helm chart and the turbinia config and manually make the change as shown below:
helm pull osdfir-charts/turbinia --untar
cd turbinia/
wget "https://raw.githubusercontent.com/google/turbinia/master/turbinia/config/turbinia_config_tmpl.py" -O turbinia.conf
# Edit the API_EVIDENCE_UPLOAD_DIR (e.g. `/mnt/turbiniavolume/uploads` )
helm install <release> ../turbinia
This might very well be a lack of understanding on my end, but I tried uploading a file to Turbinia (gpt.raw from Turbinia test-data), with turbinia-client which failed with HTTP status 500.
Looking at the server logs the exception was:
From a quick look at the
api/routes/evidence.py
it seems to reference a config variable called API_EVIDENCE_UPLOAD_DIR, which in my config is set to/evidence
. This isn't a directory as far as I can see, neither is it mounted. It seems to be the default from the config downloaded in init-configmapIt also seems that the init-configmap for Turbinia sets some values to the PVC mount in the helm chart, but for some reason MOUNT_DIR_PREFIX seems to point to another directory, which isn't the PVC mount. https://github.com/google/osdfir-infrastructure/blob/42b1774552d53091f8ae79f004037c356bd5eda6/charts/turbinia/templates/init-configmap.yaml#L56
Additional info: the stack is K3S with Longhorn, deployed with ArgoCD. I access turbinia through port-forwarding.