Self Managed Setup
Tyk’s Self-Managed setup is the deployment and orchestration of the gateway along with any other Tyk stack on your infrastructure e.g. Bare Metal, AWS AMI, Docker, Kubernetes etc.
Self Managed Setup Checklist
- Verify you are indeed calling the right endpoint on the gateway. Check the host and the port values.
-
Probe your gateway and dashboard connection. Explore whether the connection status is intact by testing via our liveliness health check endpoint. You can also dissect the gateway and dashboard interconnected config fields e.g. db_app_conf_options and tyk_api_config.
- Review your gateway configuration and validate your environment variables are not overriding the config file. If your gateway is sharded, ensure the right segment tags are added.
- Inspect the list of APIs loaded on the gateway and confirm the desired API definition is registered on the gateway. Use the gateway REST API (tyk/apis) to reveal the API definitions loaded on the gateway. If in RPC mode, peruse the gateway logs for slug binding warnings, listen_path collision errors, or duplicate API debug messages on gateway startup. You can emit logs in debug mode via the log_level field for more verbose logging.
- Enable track_404_logs on the gateway and examine the gateway logs for the proxied resource path
- Examine the latency field of the resulting analytics record. If using the dashboard, look up the logs in the Log Browser and analyse the response details. If latency is greater than zero, Tyk successfully proxied the request upstream.
- Audit your API definition configuration:
- Check if strip_listen_path is desired. Also, test if disable_strip_slash does not interrupt the final resource path.
- Investigate the port, protocol, domain, listen_path, version and internal state of API definition.
- Validate your upstream, middleware or proxy configurations. Inject a dummy upstream with lean configurations to verify there are no issues.
- If all fails, then kindly open a support request and share the response result, the request details, and any accompanying gateway logs along with your API definition.
Cloud Setup
Tyk Cloud setup is the deployment of everything you need to manage your global API ecosystem: the gateway, the dashboard, the developer portal, UDG and MDCB all in a flexible, powerful and highly scalable environment. This fully managed cloud API Management solution is available in the US, UK, Germany & Singapore.
Note: Refer to self managed setup for hybrid cloud deployments
Since major parts of the troubleshooting take place on the gateway, a self managed
setup, in this case, could also mean that you have deployed the control plane
comprised of the dashboard, MDCB and management gateway) in Tyk cloud but are using
your own gateways deployed in an environment of your choice (on-prem, Alibaba, Azure,
AWS, Cloudflare, Digital Ocean, GCP etc.)
Cloud Setup Checklist
- Verify you are indeed calling the edge gateway/ingress and not the management (control plane/dashboard) endpoint. Check the host and the port values.
- Confirm there isn't a Tyk cloud outage. Simply calling the liveliness health check endpoint should reveal the result.
- Review your plan and verify you are not exceeding your quota. Also, check that you have not enabled detailed logging on all of your APIs. Do note that it only takes one API with detailed logging enabled to swamp a gateway and cause it to OOM if that API logs a great deal of information.
- Audit your API definition configuration:
- Ensure your API definition is tagged for sharding. Check the segment tags (node segmentation) in the advanced options. In addition, there is a regression bug in all LTS version 4s before 4.0.12 where additional segment tags cause the gateway to have collision path issues. To fix this, ensure you only have one region or global tag added to the segments.
- Check if strip_listen_path is desired. Also, test if disable_strip_slash does not interrupt the final resource path.
- Investigate the port, protocol domain, listen_path, version and internal state of API definition.
- Validate your upstream, middleware or proxy configurations. Prototype or inject a dummy upstream with lean configurations to verify there are no issues.
- If all fails, then kindly open a support request and share the response result, the request details, your API definition along with the full cloud URL (screen showing the control plane and ingress endpoints) of your deployment(s) with the issue.
Comments
0 comments
Please sign in to leave a comment.