Upgrade Testing

Use this workflow to detect regressions when changing client version, config, or infrastructure.

Target workflow

Baseline run against current endpoint set
Perform upgrade
Re-run same suite(s)
Compare verdicts and call-level diffs

Prerequisites

This workflow requires a comparison suite with a stable reference endpoint that will not change during the upgrade.

Multi-endpoint suite configured with reference endpoint
Stable pre-upgrade environment
Change window for upgrade rollout

See Comparisons and Configure Suites.

Phase 1: Capture baseline

Capture a clean baseline before touching anything.

Freeze suite config (template, params, targets).
Trigger baseline run.
Record run ID and summary artifacts.

BASELINE_RUN_ID=$(curl -sS -X POST "https://api.onchainprobe.com/v1/suites/$SUITE_ID/runs" \
  -H "Authorization: Bearer $ONCHAINPROBE_API_KEY" | jq -r '.id')

Wait for completion and store:

/runs/:runId
/runs/:runId/stats
/runs/:runId/verdict

Phase 2: Apply upgrade

Apply your change while keeping the reference endpoint stable. For example:

client binary version bump
node configuration updates
infra migration (instance type, region)

Phase 3: Run post-upgrade checks

Re-run the exact same suite against the upgraded environment.

POST_RUN_ID=$(curl -sS -X POST "https://api.onchainprobe.com/v1/suites/$SUITE_ID/runs" \
  -H "Authorization: Bearer $ONCHAINPROBE_API_KEY" | jq -r '.id')

Again, wait for terminal status.

Phase 4: Compare outputs

Compare the post-upgrade results against the baseline to surface regressions.

curl -sS "https://api.onchainprobe.com/v1/runs/$POST_RUN_ID/verdict" \
  -H "Authorization: Bearer $ONCHAINPROBE_API_KEY" | jq

What to inspect:

verdict
baseline_comparison
latency percentile drift
method-specific mismatches

Example decision policy

Decide in advance what verdict outcomes mean for your rollout.

pass + acceptable latency drift: continue rollout
warn with known benign changes: continue with monitoring
fail: stop rollout and investigate before retry

Suggested evidence bundle

For auditable deployments, collect these artifacts and attach them to your change ticket.

baseline run ID
post-upgrade run ID
verdict markdown summary
top failing methods (if any)
rollback decision notes