Product logo
DocsHubBlogPricing

Common Troubleshooting Scenarios for Mods

Error: "Forbidden: Mod dist URL has expired"

Symptom

When trying to update a Type Installed control, an error message similar to "Forbidden: Mod dist URL has expired" appears. Re-running the control will not clear the error.

Cause

Guardrails distributes mods using S3 pre-signed URLs. These URLs will expire after three days. As the controls belong to the mod, the entire mod must be refreshed to get a new URL.

Remediation

  1. Login to the Guardrails Console as a user with Turbot/Owner permissions.
  2. Navigate to the Admin page by clicking the gear icon in the top right, then select the Mods tab.
  3. Find the mod that is in error in the list.
  4. Click the pencil icon. A dialog box with a number of versions will pop up. Click the version that is already installed.
  5. Click the Update Mod mod button.
  6. Guardrails will begin to download the selected version of the mod and deploy it. This can take a moment depending on the size of the environment.
  7. Once the mod re-installation is complete, the Mod Health control will re-run. If the control does not go into the OK state, contact the Turbot Guardrails support team at help@turbot.com.

Mod Management in Large Environments

Mod installation in large environments requires care and attention. For the purposes of this document, we will define a large environment as having more than 100 active AWS accounts. Failure to follow these processes may result in an unhealthy workspace, excessive controls in error, delayed processing and/or high operating costs during the change. The AWS Event Handler steps don't apply when installing or removing Azure and GCP mods.

New Mod Installation Process

  1. Identify the list of new mods to be installed.
  2. Set AWS > Turbot > Event Handlers to Skip. We don't want to kick off any Event Handler Terraform runs until we are ready. The Skip setting tells Guardrails to ignore any changes but preserve any existing Event Handler infrastructure.
  3. Install the new mod. Avoid installing more than one mod at a time.
  4. Wait until all new Discovery and CMDB controls have transitioned out of tbd. Letting TBDs accumulate over several mod installations will complicate and extend clean up. All Discovery and CMDB controls should be in ok before installing the next mod.
  5. Repeat steps 3 and 4 until all new mods have been installed.
  6. Resolve workspace issues related to the mod installs:
    • Identify the cause of any persistent policy values in error or invalid.
    • If you have large numbers of policy values in tbd and the number is not steadily decreasing over time, resolve that issue before tackling any controls in tbd. (Controls depend on policy values to operate. Controls with policy values in tbd cannot transition out of tbd themselves.)
    • Identify any issues with controls in tbd, error or invalid. Resolve any problems.
    • Use the run_policy and run_controls scripts help a lot with resolving policy values and controls stuck in tbd.
    • A healthy workspace has zero controls or policy values in error or invalid. A small number of short-lived controls/policy values in tbd is okay.
  7. Set AWS > Turbot > Event Handlers to Enforce: Configured. Behind the scenes all the event handler source policies will have updated to include the new Event Rules. We want the event handlers to run once, not every time each mod is installed.
  8. Event Handlers are good to go when all Event Handler, Event Rule > Configured and Event Target > Configured controls are in ok.
  9. (Optional) It may be desirable to rerun the Discovery and CMDB controls for the new mods. A resource may have change between when the Discovery control ran and when the event handlers updated to handle the new events.

Mod Upgrade Process

It's best practice to set the Turbot > Mod > Auto Update > Schedule policy to run mod updates on the weekends. (Turbot Guardrails SaaS customers are already configured to use an Auto Update schedule.) This way, if a mod upgrade does generate event handler churn, it's during off hours.

Mod Removal Process

Mod removal is most reliable when all the resources for that mod have been removed from Guardrails first. For Azure and GCP mod removal, the AWS Event Handler steps should be disregarded.

  1. Identify the list of mods to remove.
  2. For each mod, identify the list of resources managed by that mod. Reading the mods docs (/guardrails/docs/mods/) can help.
  3. For your workspace identify how many resources of those resource types exist.
  4. (Optional) build a Terraform template that will set all the CMDB policies for those resources to Enforce: Disabled
  5. Set AWS > Turbot > Event Handlers to Skip across all resources. (We set these here to prevent churn from the aws-ec2 and aws-vpc-* mods where event handlers are calculated on a per-resource basis.)
  6. Apply the CMDB policies at the Turbot level manually (or via terraform) in batches start with the resource types with the lowest total number of resources and move up to the higher volume resources to not overwhelm the database.
  7. Let Guardrails clean up the resources. Time taken will depend on overall system load, Guardrails configuration and the number of resources for that mod.
    • Use the run_controls script in the case where CMDB controls need a kick to finish cleaning themselves up.
  8. Verify that all resources for each mod have been removed from Guardrails. There should be zero resources for each mod at this point in the process.
  9. Remove each mod from the admin page.
  10. Wait 20 minutes for the Event Handler source policies to recalculate.
  11. Set AWS > Turbot > Event Handlers to Enforce: Configured.
  12. Verify that Event Handler controls are in ok.

Using Guardrails Interval Policies

Guardrails, by default, utilizes an event driven model to detect resource changes. In some rare circumstances this is either not possible or does not meet very specific business needs.

This introduces us to the concept of Guardrails Intervals. Intervals can be used to force regular “ticking” of CMDB controls, which then trigger guardrails when misconfigurations are detected. While this method can solve many use cases, it is important to recognize the cost associated with doing so.

Why Intervals?

  • Ability to regularly update resource metadata which generate no events when changed, i.e. the number of available IP addresses in a subnet.
  • An organization requires resources to be scanned and logged on a regular basis.

Guardrails Intervals come at a cost. Each “tick” represents a request from Guardrails to describe a set of resources. As the number of resources in the environment goes up, so does the amount of requests sent to and received from the cloud service API endpoint. Similarly, setting a short interval triggers more requests.

Setting a Guardrails Interval

In order to set a Guardrails Interval policy, you must have Turbot/Admin permissions at the root level, CLI configured, and Terraform configured.

1. Either create a new Terraform configuration file or open an existing one in a text editor.

2. Paste the following code in the configuration file:

resource "turbot_policy_setting" "turbot_interval" {
  resource = "tmod:@turbot/aws-ec2#/control/types/volumeCmdb"
  type     = "tmod:@turbot/turbot#/policy/types/interval"
  value    = "days: 1"
  note     = "[Ticket CLOUD-152] Run the volume CMDB control on a scheduled interval"
}

Let's break this down:

  • resource - This is the resource where the policy will be set. Notice it is a Control for Volume CMDB. We can use any control here that needs to be triggered on a regular interval. For example, if the requirement was to refresh S3 Bucket CMDB entries, the control would then be tmod:@turbot/aws-s3#/control/types/s3AccountCmdb. Use the Mods Registry to find the correct control
  • type - The policy type is defining which policy is being created. This URI corresponds to the policy Turbot > Interval.
  • value - This defines the time in between control runs. It is recommended to start at the upper bound of the time requirements and slowly increment towards the desired time requirement. Additional keys that can be used are hours and minutes. Great care must be taken to not set too small of an interval!
  • note - While not required, it is highly recommended to set a note for future you (and anyone viewing said policy)!

3. Apply the configuration file.

4. That's it! CMDB controls will now be triggered for the desired resource type (EC2 volumes for this example) on the defined interval.

Important Note: This works well for new controls, but existing controls will require a control run to adhere to the new, defined interval. We suggest running all relevant resource controls which can be done all at once by utilizing a control run script.

WARNING!!

Cloud providers can and will throttle requests once the request rate is exceeded. Regular control runs at short intervals can also generate a considerable amount of “noise” in the environment.

Mod Interval policies are applied on controls themselves and apply for ALL resources defined by the control. For example, setting the policy Turbot > Interval policy on the tmod:@turbot/aws-ec2#/control/types/volumeCmdb control type will cause Guardrails to poll ALL EC2 volumes in ALL accounts imported into Guardrails. This CANNOT be set at a lower level in the hierarchy!

If setting a Guardrails Mod Interval policy is required, it is recommended to start at the upper bound of the time frame. For example, start with 24 hours per tick to determine the impact on your environment and adjust downwards as necessary and as the environment allows it.

Guardrails is not liable for misconfigurations of the Interval policy resulting in application downtime and throttling of service API. Great care must be taken when setting the interval to not overload the API endpoints.