API timeout running on large repository

[alclonky]

We are getting timeouts calling /status or / with an liveness probe when renovate runs for an large repo
this repos has a renovate run duration of 15 minutes

    "stats": {
         "managers": {
           "dockerfile": {"fileCount": 6, "depCount": 6},
           "gradle": {"fileCount": 69, "depCount": 16560},
           "html": {"fileCount": 3, "depCount": 6},
           "npm": {"fileCount": 3, "depCount": 22},
           "nuget": {"fileCount": 2, "depCount": 0}
         },
         "total": {"fileCount": 83, "depCount": 16594}

We are running renovate 1.0.0-rc.3 on an OpenShift 4.4 Cluster
CPU Usage is between 700 and 800m, Memory Usage around 8 GB

if I look at the logs during such timeout situation the run seems "stuck"

[rarkins]

this repos has a renovate run duration of 15 minutes

From other conversations we had by email, I think you mean you cron the scheduled task every 15 minutes? That doesn't actually limit the duration.

Does this repository eventually finish?

[alclonky]

No thats the duration it need until finished and also the cron

[Reamer]

We also see dropouts of the API while the cron checks repositories.
Maybe a dedicated thread which responds to API requests will help here.

[Reamer]

We have increased our timeout from one second from the Helm File to ten seconds. So far, no other errors have appeared.

...
      readinessProbe:
        httpGet:
          path: /
          port: renovate
          scheme: HTTP
        timeoutSeconds: 10
        periodSeconds: 30
        successThreshold: 1
        failureThreshold: 3
...
      livenessProbe:
        httpGet:
          path: /
          port: renovate
          scheme: HTTP
        initialDelaySeconds: 30
        timeoutSeconds: 10
        periodSeconds: 30
        successThreshold: 1
        failureThreshold: 3
...