repo-server prints Invalid log messages on liveness check failures

[dudinea]

Checklist:

• I've searched in the docs and FAQ for my answer: https://bit.ly/argocd-faq.
• I've included steps to reproduce the bug.
• I've pasted the output of argocd version.

Describe the bug

When repo-server liveness check (/healthz?full=true) fails, it prints an invalid and confusing error log message like:

time="2026-01-11T22:34:04Z" level=error msg="&{0xc000884bd0 0xc001310640 {} 0x4fba00 true false true {{} {0 0}} {{} 0} 0xc0001033c0 {0xc0013163c0 map[] false false} map[] false 0 -1 503 false false false [] {{} 0} [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0] [0 0 0 0 0 0 0 0 0 0] [0 0 0] {{} {0 0}} <nil> true} rpc error: code = Canceled desc = context canceled"

Originally reported by @maximemoreillon in #25931 . This issue is opened to fix this specific problem with logging, not the original issue.

To Reproduce

Make the liveness check fail to respond in time.

One way to do that is to configure in the deployment:

• liveness check timeout to 1 second (the min. value possible)
• give to the repo-server pod an inadequately small cpu limit:

        livenessProbe:
          timeoutSeconds: 1
...
        resources:
          limits:
            cpu: 1m

The repo server will start and it will fail to respond to the liveness
probe under the specified 1 second timeout, then kubernetes will
close the HTTP connection to the pod, which will cause to print
the incorrect error message.

Expected behavior

A correct and useful message should be printed

Screenshots

Version

argocd: v3.2.3+2b6251d
  BuildDate: 2025-12-24T12:35:36Z
  GitCommit: 2b6251dfedb54de40596272a73ed1fb19d740219
  GitTreeState: clean
  GoVersion: go1.25.0
  Compiler: gc
  Platform: windows/amd64