mengw15 opened a new issue, #6332:
URL: https://github.com/apache/texera/issues/6332

   ### What happened?
   
   `.github/scripts/smoke-boot.sh` greps the **entire** service boot log for a 
crash regex to detect runtime linkage failures ([line 
~100](https://github.com/apache/texera/blob/main/.github/scripts/smoke-boot.sh)):
   
   ```sh
   
crash_re='NoClassDefFoundError|ClassNotFoundException|LinkageError|NoSuchMethodError|...|requires
 Jackson Databind'
   if grep -qE "$crash_re" "$log"; then
     fail "'$launcher' hit a runtime classpath/linkage error on boot"
   fi
   ```
   
   jOOQ prints a **random "tip of the day"** banner at startup. One of those 
tips reads: *"A NoClassDefFoundError or ClassNotFoundException is often a sign 
that your jOOQ code is generated with a different version of jOOQ…"* — 
informational prose that contains `NoClassDefFoundError` and 
`ClassNotFoundException`, so `crash_re` matches it and the check fails **even 
though the service booted fine** (Jetty opened its port).
   
   Because jOOQ picks the tip at random, this is **flaky**: it fails only when 
that particular tip is drawn, on whichever service drew it. Every platform 
service uses jOOQ (via DAO), so any service on any PR can randomly hit it.
   
   **Observed:** `platform-integration (config-service)` on #6331 — a PR that 
only added a test file, unrelated to config-service. The boot log shows `Opened 
application@…{0.0.0.0:9094}` (booted successfully), yet the check failed on the 
jOOQ tip line.
   
   **Impact:** random red CI on unrelated PRs; blocks merges until a re-run 
happens to draw a different tip.
   
   **Scope:** `smoke-boot.sh` is the single shared crash-detection script — all 
boot checks call it (currently the `platform-integration` matrix, 
`build.yml:889`), so it affects every caller. No other CI step does log-grep 
crash detection.
   
   ### How to reproduce?
   
   1. Run `smoke-boot.sh` against a service whose boot log contains jOOQ's 
`NoClassDefFoundError`/`ClassNotFoundException` tip (drawn at random on 
startup) — the check fails despite a clean boot.
   2. Concrete instance: #6331 CI, `platform-integration (config-service)` leg 
— service opened port 9094, but the step failed on the jOOQ tip.
   
   ### Version
   
   1.1.0-incubating (Pre-release/Master)


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to