This topic describes troubleshooting methods you can use with Supply Chain Security Tools (SCST) - Scan.
Run these commands to get more logs and details about the errors around scanning. The TaskRuns and pods persist for a predefined amount of seconds before getting deleted. (deleteScanJobsSecondsAfterFinished
is the tap pkg variable that defines this)
To retrieve TaskRun events:
kubectl describe taskrun TASKRUN-NAME -n DEV-NAMESPACE
Where:
TASKRUN-NAME
is the name of the TaskRun.DEV-NAMESPACE
is the name of the developer namespace you want to use.Run the following to get error logs from a pod when scan pods are in a failing state:
kubectl logs scan-pod-name -n DEV-NAMESPACE
Where DEV-NAMESPACE
is the name of the developer namespace you want to use.
See here for more details about debugging Kubernetes pods.
The following is an example of a successful scan run output:
scan:
cveCount:
critical: 20
high: 120
medium: 114
low: 9
unknown: 0
scanner:
name: Grype
vendor: Anchore
version: v0.37.0
reports:
- /workspace/scan.xml
eval:
violations:
- CVE node-fetch GHSA-w7rc-rwvf-8q5r Low
store:
locations:
- https://metadata-store-app.metadata-store.svc.cluster.local:8443/api/sources?repo=hound&sha=5805c6502976c10f5529e7f7aeb0af0c370c0354&org=houndci
A scan run that has an error means that one of the init containers: scan-plugin
, metadata-store-plugin
, compliance-plugin
, summary
, or any other additional containers had a failure.
To inspect for a specific init container in a pod:
kubectl logs scan-pod-name -n DEV-NAMESPACE -c init-container-name
Where DEV-NAMESPACE
is the name of the developer namespace you want to use.
See Debug Init Containers in the Kubernetes documentation for debug init container tips.
To retrieve status conditions of an SourceScan and ImageScan, run:
kubectl describe sourcescan SOURCE-SCAN -n DEV-NAMESPACE
Where:
DEV-NAMESPACE
is the name of the developer namespace you want to use.SOURCE-SCAN
is the name of the SourceScan you want to use.kubectl describe imagescan IMAGE-SCAN -n DEV-NAMESPACE
Where:
DEV-NAMESPACE
is the name of the developer namespace you want to use.IMAGE-SCAN
is the name of the ImageScan you want to use.Under Status.Conditions
, for a condition look at the “Reason”, “Type”, “Message” values that use the keyword “Error” to investigate issues.
See here for Tanzu workload commands for tailing build and runtime logs and getting workload status and details.
To retrieve scan-controller manager logs:
kubectl -n scan-link-system logs -f deployment/scan-link-controller-manager -c manager
If you encounter an issue with the scan-link controller not starting, run the following to restart the deployment to see if it’s reproducible or flaking upon starting:
kubectl rollout restart deployment scan-link-controller-manager -n scan-link-system
If you encounter this issue:
✖ Error: Post "https://metadata-store.tap.tanzu.example.com/api/sourceReport?": \
tls: failed to verify certificate: x509: certificate signed by unknown authority
To ensure that the caSecret
from the scanner DEV-NAMESPACE
matches the caSecret
from the METADATASTORE-NAMESPACE
namespace:
In a single cluster, the connection between the scanning pod and the metadata store happens inside the cluster and does not pass through ingress. This is automatically configured. You do not need to provide an ingress connection to the store. If you provided an ingress connection to the store, delete it.
Get the caSecret.name
. If you are using a multicluster setup, retrieve grype.metadataStore.caSecret.name
from the Grype config:
grype:
metadataStore:
caSecret:
name: store-ca-cert
importFromNamespace: metadata-store-secrets
Note
caSecret.name
is set tostore-ca-cert
. For more information, see Set up multicluster Supply Chain Security Tools (SCST) - Store.
If you are using a single-cluster setup, the default value for grype.metadataStore.caSecret.name
is app-tls-cert
. For more information, see Install Supply Chain Security Tools - Scan.
Verify that the CA-SECRET
secret exists in the DEV-NAMESPACE
by runnning:
kubectl get secret CA-SECRET -n DEV-NAMESPACE
If the secret CA-SECRET
doesn’t exist in your DEV-NAMESPACE
, verify that the CA-SECRET
exists in the METADATASTORE-NAMESPACE
namespace:
kubectl get secret CA-SECRET -n METADATASTORE-NAMESPACE
Where METADATASTORE-NAMESPACE
is the namespace that contains the secret CA-SECRET
.
If you are using a single cluster, it is configured using the metadata-store
namespace. If multicluster, it is configured using the metadata-store-secrets
.
If CA-SECRET
does not exist in the metadata store namespace, configure the certificate. See Custom certificate configuration.
Check if the secretexport and secretimport exist and are reconciling successfully by running:
kubectl get secretexports.secretgen.carvel.dev -n `METADATASTORE-NAMESPACE`
kubectl get secretimports.secretgen.carvel.dev -n `DEV-NAMESPACE`
SCST - Store creates the single cluster secretexport by default. See Deployment details and configuration. For information about creating the multicluster secretexport, see Set up multicluster Supply Chain Security Tools (SCST) - Store.
Verify that the ca.crt
field in both secrets from METADATASTORE-NAMESPACE
and DEV-NAMESPACE
match, or that the ca.crt
field of the secret in the METADATASTORE-NAMESPACE
includes the ca.crt
field of the DEV-NAMESPACE
secret.
Confirm this by base64 decoding both secrets and verifying that there is a match:
kubectl get secret CA-SECRET -n DEV-NAMESPACE -o json | jq -r '.data."ca.crt"' | base64 -d
kubectl get secret CA-SECRET -n METADATASTORE-NAMESPACE -o json | jq -r '.data."ca.crt"' | base64 -d
The certificates in the METADATASTORE-NAMESPACE
and DEV-NAMESPACE
must have a match for the scanner to connect to the metadata-store.
The source scan step is opt-in in Tanzu Application Platform v1.6 to better support languages that resolve dependencies at build time. For information and how to opt-in to source scanning in the out-of-the-box test and scan supply chain, see Scan Types for Supply Chain Security Tools - Scan.
For information about issues with Grype in air gap environments, see Use vulnerability scanning in offline and air-gapped environments.
Scanning source code from a private source repository requires an SSH secret present in the namespace and referenced as grype.targetSourceSshSecret
in tap-values.yaml
. See Installing the Tanzu Application Platform Package and Profiles.
If a private source scan is triggered and the secret cannot be found, the scan pod includes a FailedMount
warning in Events with the message:
MountVolume.SetUp failed for volume "ssh-secret" : secret "secret-ssh-auth" not found
Where secret-ssh-auth
is the value specified in grype.targetSourceSshSecret
Scanning an image from a private registry requires an image pull secret to exist in the Scan CRs namespace and be referenced as grype.targetImagePullSecret
in tap-values.yaml
. See Installing the Tanzu Application Platform Package and Profiles.
If a private image scan is triggered and the secret is not configured, the scan TaskRun’s pod’s step-scan-plugin
container fails with the following error:
Error: GET https://dev.registry.tanzu.vmware.com/v2/vse-dev/spring-petclinic/manifests/sha256:128e38c1d3f10401a595c253743bee343967c81e8f22b94e30b2ab8292b3973f: \
UNAUTHORIZED: unauthorized to access repository: vse-dev/spring-petclinic, action: pull: unauthorized to access repository: vse-dev/spring-petclinic, action: pull
SCST - Store is required to install SCST - Scan.
If you install without the SCST - Store, you must edit the configurations to deactivate the Store:
---
metadataStore:
url: ""
Install the package with the edited configurations by running:
tanzu package install scan-controller \
--package scanning.apps.tanzu.vmware.com \
--version VERSION \
--namespace tap-install \
--values-file tap-values.yaml
You might encounter the following error:
The provided SBOM has a Syft Schema Version which doesn't match the version that is supported by Grype...
This means that the Syft Schema Version from the provided SBOM doesn’t match the version supported by the installed grype-scanner
. There are two different methods to resolve this incompatibility issue:
Install a version of Tanzu Build Service that provides an SBOM with a compatible Syft Schema Version. This is the method VMware recommends.
Deactivate the failOnSchemaErrors
in grype-values.yaml
. See Install Supply Chain Security Tools - Scan. Although this change bypasses the check on Syft Schema Version, it does not resolve the incompatibility issue and produces a partial scanning result.
syft:
failOnSchemaErrors: false
If your scan policy appears to not be enforced, it might be because the Rego file defined in the scan policy is incompatible with the scanner that is being used. For example, the Grype Scanner outputs in the CycloneDX XML format while the Snyk Scanner outputs SPDX JSON.
See Sample ScanPolicy for Snyk in SPDX JSON format for an example of a ScanPolicy formatted for SPDX JSON.
If you encounter the following issue, it might be due to not exporting app-tls-cert
to the correct namespace:
{"level":"error","ts":"2022-06-08T15:20:48.43237873Z","logger":"setup","msg":"Could not find CA in Secret","err":"unable to set up connection to Supply Chain Security Tools - Store"}
Configure ns_for_export_app_cert
in your tap-values.yaml
file.
metadata_store:
ns_for_export_app_cert: "DEV-NAMESPACE"
Where DEV-NAMESPACE
is the name of the developer namespace you want to use.
If there are multiple developer namespaces, use ns_for_export_app_cert: "*"
.
A Source Scan for a blob artifact can cause reporting in the status.artifact
and status.compliantArtifact
the wrong URL for the resource, passing the remote SSH URL instead of the cluster local fluxcd one. One symptom of this issue is the image-builder
failing with a ssh:// is an unsupported protocol
error message.
You can confirm you’re having this problem by running kubectl describe
in the affected resource and comparing the spec.blob.url
value against the status.artifact.blob.url
. The problem occurs if they are different URLs. For example:
kubectl describe sourcescan SOURCE-SCAN-NAME -n DEV-NAMESPACE
Where:
SOURCE-SCAN-NAME
is the name of the source scan you want to configure.DEV-NAMESPACE
is the name of the developer namespace you want to use.And compare the output:
...
spec:
blob:
...
url: http://source-controller.flux-system.svc.cluster.local./gitrepository/sample/repo/8d4cea98b0fa9e0112d58414099d0229f190f7f1.tar.gz
...
status:
artifact:
blob:
...
url: ssh://[email protected]:sample/repo.git
compliantArtifact:
blob:
...
url: ssh://[email protected]:sample/repo.git
This problem happens in SCST - Scan v1.2.0
when you use a Grype Scanner ScanTemplates earlier than v1.2.0
, because this is a deprecated path. To fix this problem, upgrade your Grype Scanner deployment to v1.2.0
or later. See Upgrading Supply Chain Security Tools - Scan for step-by-step instructions.
If the Supply Chain is not progressing due to CVEs found in either the SourceScan or ImageScan, see the CVE triage workflow in Triaging and Remediating CVEs.
If you encounter No policy has been defined
, it might be because the Tanzu Application Platform GUI is unable to view the Scan Policy resource.
Confirm that the Scan Policy associated with a SourceScan or ImageScan exists. For example, the scanPolicy
in the scan matches the name of the Scan Policy.
kubectl describe sourcescan NAME -n DEV-NAMESPACE
kubectl describe imagescan NAME -n DEV-NAMESPACE
kubectl get scanpolicy NAME -n DEV-NAMESPACE
Where DEV-NAMESPACE
is the name of the developer namespace you want to use.
Add the app.kubernetes.io/part-of
label to the Scan Policy. See Enable Tanzu Application Platform GUI to view ScanPolicy Resource.
If your scan pod is failing, you might see the following connection error in the logs:
dial tcp: lookup metadata-store-app.metadata-store.svc.cluster.local on 10.100.0.10:53: no such host
A connection error while attempting to connect to the local cluster URL causes this error. If this is a multicluster deployment, set the grype.metadataStore.url
property in your Build profile values.yaml
. You must set the ingress domain of SCST - Store which is deployed in the View cluster. For information about this configuration, see How to configure Grype in the Build profile values file.
If your Source Scan resource is failing, the status might show this error:
Error: endpoint require 'http://' or 'https://' prefix
This is because the grype.metadataStore.url
value in the Tanzu Application Platform profile values.yaml
was not configured with the correct prefix. Verify that the URL starts with either http://
or https://
.
If the scan phase is in Error
and the status condition message is:
Summary logs could not be retrieved: . error opening stream pod logs reader: container summary is \
not valid for pod scan-grypeimagescan-sample-public-zmj2g-hqv5g
This error might be a consequence of using Grype Scanner ScanTemplates shipped with SCST - Scan v1.1 or earlier. These ScanTemplates are deprecated and are not supported in Tanzu Application Platform v1.4.0 and later.
There are two options to resolve this issue:
Option 1: Upgrade to the latest Grype Scanner version. This automatically replaces the old ScanTemplates with the upgraded ScanTemplates.
Option 2: Create a ScanTemplate. Follow the steps in Create a scan template.
The following error in the pod logs indicate that the self-signed certificate might be incorrectly configured:
x509: certificate signed by unknown authority
To resolve this issue, ensure that shared.ca_cert_data
contains the required certificate. For an example of setting up the shared self-signed certificate, see Build profile.
For information about shared.ca_cert_data
, see View possible configuration settings for your package.
The docker
field and related sub-fields by SCST - Scan Controller, Grype Scanner, or Snyk Scanner are deprecated in Tanzu Application Platform v1.4.0 and removed in Tanzu Application Platform v1.10.1. Previously these fields might be used to populate the registry-credentials
secret. You might encounter the following error during installation:
UNAUTHORIZED: unauthorized to access repository
The recommended migration path for users setting up their namespaces manually is to add registry credentials to both the developer namespace and the scan-link-system
namespace, using these instructions.
ImportantThis step does not apply to users who used
--export-to-all-namespaces
when setting up the Tanzu Application Platform package repository.
Before running a scan, the Grype scanner downloads a copy of its database. If the database fails to download, the following log entry might appear.
Vulnerability DB [no update available] New version of grype is available: 0.50.2 [0000] WARN unable \
to check for vulnerability database update 1 error occurred: * failed to load vulnerability db: \
vulnerability database is corrupt (run db update to correct): database metadata not found: \
~/Library/Caches/grype/db/3
To resolve this issue, ensure that Grype has access to its vulnerability database:
This issue is unrelated to Supply Chain Security Tools for Tanzu – Store.
v1.5.0
or laterFor SCST - Scan v1.5.0
or later, you see scanner pods restart:
Pods
NAME READY STATUS RESTARTS AGE
my-scan-45smk-pod 0/9 Completed 1 14m
One restart in scanner pods is expected with successful scans. To support Tanzu Service Mesh (TSM) integration, jobs were replaced with TaskRuns. This restart is an artifact of how Tekton cleans up sidecar containers by patching the container specifications.
When upgrading the SCST - Scan from a previous version to v1.10, you might see a reconciliation failure similar to:
NAME PACKAGE-NAME PACKAGE-VERSION STATUS
scanning scanning.apps.tanzu.vmware.com 1.10.1-build.36081392+c072d305 Reconcile failed
If getting the package by running the command tanzu package installed get scanning -n tap-install
you might see an error similar to:
STATUS: Reconcile failed
CONDITIONS: - type: ReconcileFailed
status: "True"
reason: ""
message: Error (see .status.usefulErrorMessage for details)
USEFUL-ERROR-MESSAGE: ytt: Error: Overlaying data values (in following order: additional data values):
One or more data values were invalid
====================================Given data value is not declared in schema
values.yaml:
|
2 | url: ""
| = found: url
= expected: a map item with the key named "exports" (from .ytt/data.yaml:52)
This is because the field scanning.metadataStore.url
is removed. If this field is present in tap-values.yaml
provided for the upgrade, the reconciliation fails. To resolve this problem, remove the field from the values file and run the upgrade command again.
As part of compliance with the restricted profile Kubernetes Pod Security Standards, you must set the securityContext
of containers and initContainers. This applies to the prepare
initContainers created by Tekton. When a pod does not meet pod Security Standards, it is not created and vulnerability scanning cannot proceed. For more information, see the Kubernetes documentation.
You might see an error message similar to the following when describing the TaskRun:
"scan-source-scan-with-passing-policy-zx46t-pod" is forbidden: violates PodSecurity "restricted:latest": \
allowPrivilegeEscalation != false (container "prepare" must set securityContext.allowPrivilegeEscalation=false), \
unrestricted capabilities (container "prepare" must set securityContext.capabilities.drop=["ALL"]), \
seccompProfile (pod or container "prepare" must set securityContext.seccompProfile.type to \
"RuntimeDefault" or "Localhost"). Maybe invalid TaskSpec. ScanPodError PodNotFound: no pod found
Update your Tekton Pipelines package configuration in your tap-values.yaml
file with the following changes.
tekton_pipelines:
feature_flags:
set_security_context: "true"
Setting the securityContext
resolves the prepare
initContainer violation.
Update your Tanzu Application Platform installation by running:
tanzu package installed update tap -p tap.tanzu.vmware.com -v TAP-VERSION --values-file tap-values.yaml -n tap-install
Where TAP-VERSION
is the version of Tanzu Application Platform installed.
Re-run the scan.