Skip to content

chore(prod): dinersclub v0.0.46 - #153

Merged
nandanrao merged 1 commit into
mainfrom
chore/deploy-dinersclub-046-prod
Aug 20, 2026
Merged

chore(prod): dinersclub v0.0.46#153
nandanrao merged 1 commit into
mainfrom
chore/deploy-dinersclub-046-prod

Conversation

@nandanrao

Copy link
Copy Markdown
Contributor

Promotes #151 / #152 to production. Staging has run v0.0.46 since helm rev 81.

Staging evidence

restarts 0
scrape target serviceMonitor/vstag/gbv-dinersclub/0 — up, no error
real payments 1 success + 1 COULD_NOT_AUTO_DETECT_OPERATOR (classified permanent, delivered to the respondent as before)
unclassified codes none
processing faults none
payout latency ~1.5–1.8s, i.e. the permanent failure was not retried — it would have burned the full 60s budget if misclassified

Blast radius: one image

Checked main's production.yaml against all 27 gbv-owned workloads in vprod before bumping — already identical. So:

gbv-dinersclub/dinersclub: v0.0.45 -> v0.0.46

and nothing else; no new workloads. That silent-passenger problem is exactly what PR #147 was written about.

PR #147 should be closed, not merged

It pins dinersclub back to v0.0.43-wa because v0.0.45 moves DingConnect to Generic Secrets. That's stale: v0.0.45 has since been deployed on its own, the v0.0.45..v0.0.46 range doesn't touch dingconnect.go, and dingconnect isn't in DINERSCLUB_PROVIDERS in either environment.

Alert rules land after this

vlab-alerts is deployed after the image, because DinersClubMetricsMissing is legitimately true until prod is scraped. That deploy is purely additive — verified payment-health is the only rule added, with none changed or removed.

🤖 Generated with Claude Code

https://claude.ai/code/session_016w2xUDzsqh6VJei1UgmgPg

Staging has run v0.0.46 since helm rev 81 with zero restarts, a healthy scrape
target, and two real Reloadly payments through it: one success, one
COULD_NOT_AUTO_DETECT_OPERATOR classified permanent and delivered to the
respondent exactly as before. No unclassified codes, no processing faults.

Verified before applying: main's production.yaml already matched live vprod on
every one of the 27 gbv-owned workloads, so this upgrade changes exactly one
image and adds no new workload. That was the failure mode PR #147 was written
about -- a version sitting ahead of the cluster becomes a silent passenger on
the next unrelated `helm upgrade`.

PR #147 itself is stale and should be closed rather than merged: it pins
dinersclub back to v0.0.43-wa on the grounds that v0.0.45 switches DingConnect
to Generic Secrets, but v0.0.45 has since been deployed on its own, and the
v0.0.45..v0.0.46 range does not touch dingconnect.go at all. dingconnect is
also absent from DINERSCLUB_PROVIDERS in both environments.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016w2xUDzsqh6VJei1UgmgPg
@netlify

netlify Bot commented Aug 20, 2026

Copy link
Copy Markdown

Deploy Preview for vlab-research canceled.

Name Link
🔨 Latest commit 732c1dc
🔍 Latest deploy log https://app.netlify.com/projects/vlab-research/deploys/6a877635c45696000824132e

@netlify

netlify Bot commented Aug 20, 2026

Copy link
Copy Markdown

Deploy Preview for virtuallab-videos canceled.

Name Link
🔨 Latest commit 732c1dc
🔍 Latest deploy log https://app.netlify.com/projects/virtuallab-videos/deploys/6a877635b49b32000887d243

@nandanrao
nandanrao merged commit 7ed7dff into main Aug 20, 2026
4 checks passed
@nandanrao
nandanrao deleted the chore/deploy-dinersclub-046-prod branch August 20, 2026 22:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant