Please enable JavaScript.

Coggle requires JavaScript to display documents.

Mulesoft / TP Training (Day 5 - Chaos Monkey (Environment Details (mule 9,…

- - - - stateless
      - any state held on the server
    - - System API
        
        systems of record
        
        transform data
        
        connectivity - HTTP + JSON
        
        raml contract
      - Process API
        
        business functionality
        
        transform data
        
        connectivity - HTTP + JSON
        
        orchestration
        
        raml contract
      - Experience API
        
        user experience based
        
        orchestration
        
        transform
        
        connectivity - HTTP + JSON
      - RAML - Restful Architecture Modeling Language
        
        based on yaml
        
        defines the API contract
        
        uses JSON response
        
        designed for re-use
      - backlog item: create a best practice / guidance paper on use of API led architecture
    - - backlog item: platform healthcheck app is not in use equally across all regions
      - backlog item: there is an internal TP (INT) set of reverse proxies and gateways used for staff facing services (mostly CMB). Need to identify runbooks, dashboards for internal platform.
      - backlog item: have cross data center connectivity in HK but not in UK/US - why? According to Ash there is a reason, we need to follow that up.
      - Client > GSM > F5 > Reverse Proxy > F5 > Gateway Stripe > GSM > F5 > Go Router > PCF
      - multiple healthchecks
        
        platform healthcehck app
        
        gateway proxy healthcheck from F5
        
        TCP healthcheck from F5 to reverse proxy
        
        goRouter TCP check per PCF app
      - Gateway Proxy Processing: HTTP listener > API Kit > Policy Execution > HTTP > PCF. You can also apply policies on the way out
      - backlog item: we have lots of DSDs for manual configuration of security policies within PCE - why is it manual?
      - backlog item: need rate limiting guidelines
- - - - access logs
      - error logs
  - - - worker threads handle all apikit, policy execution, and http out processing.
- - - - Common Errors: 500, 502 - bad gateway, 503 - service unavailable, 403, 429
      - backlog: we need to check if the apache extension is fully applied across the TP estate (internal/external)
      - infrastructure monitoring
        
        appd machine agent
        
        apache extension??
        
        script (python) runs (via cron) grabbing infra metrics and feeding to splunk
    - - mule agent extension healthcheck - /mule/health; called from iDMZ F5
      - backlog: our entry point (DNS) management is only one rbwm-api-[region]. However there are some consumers who have defined thier own GSS entry points which we do not manage.
      - backlog: team has noticed that appd infra metrics do not seem accurate when compared to metrics gathered from the box. e.g. cpu load
    - - lots of issues in the GW layer stem from the HSM
        
        firmware bug where HSM thinks its overheating
  - - - if intermittent, likely down to a specific node or nodes, potentially down to saas or hsms
- - - - no more rewrite rules at the rProxy
- - - - client receiving a 403 intermittently due to saml signing error
        
        check on gateway logs showed nfast hardserver error
        
        HSM issue prevents saml policy from executing
        
        solution is to restore the hardserver process and restart mule runtime
  - - - rProxy was returning error 429
        
        The HTTP 429 Too Many Requests response status code indicates the user has sent too many requests in a given amount of time ("rate limiting")
        
        checked PCE - API definition config and saw Rate Limit Policy was set to 1 request per Min