From e8c927921420b7485b2267ecbaec0c79d58dfdbb Mon Sep 17 00:00:00 2001
From: Owen Nelson <onelson@svix.com>
Date: Mon, 11 Nov 2024 12:56:22 -0800
Subject: [PATCH] Bridge: add a test to check process liveness

Mentioned in the script comments, we've seen segfaults in bridge coming
from deno, and only after the interpreter has had some time to achieve
steady-state.

This diff adds a script to run a check where we launch the process with
an empty config (no senders or receivers registered), and send it to the
background.
Bridge should be able to stay up indefinitely like this. If the process ends
before we check on it (after sleeping a spell), that's probably a bug.

I revered the last deno downgrade to try this script out in the failing
state. It seemed to catch the issue well.

```
$ ./run-idle-test.sh
   Compiling svix-bridge v1.39.0 (/home/onelson/Projects/svix-webhooks/bridge/svix-bridge)
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 1.29s
Monitoring PID=189293
fail: process terminated prematurely
```
---
 .github/workflows/bridge-ci.yml |  5 +++++
 bridge/run-idle-test.sh         | 18 ++++++++++++++++++
 2 files changed, 23 insertions(+)
 create mode 100755 bridge/run-idle-test.sh

diff --git a/.github/workflows/bridge-ci.yml b/.github/workflows/bridge-ci.yml
index df8caef59..89655a5ec 100644
--- a/.github/workflows/bridge-ci.yml
+++ b/.github/workflows/bridge-ci.yml
@@ -88,6 +88,11 @@ jobs:
     - name: Stop required services
       run: docker compose -f "bridge/testing-docker-compose.yml" down
 
+    - name: Idle Test
+      working-directory: bridge
+      run: ./run-idle-test.sh
+
+
 #  deny-check:
 #    name: cargo-deny check
 #    runs-on: ubuntu-24.04
diff --git a/bridge/run-idle-test.sh b/bridge/run-idle-test.sh
new file mode 100755
index 000000000..c4daf71c1
--- /dev/null
+++ b/bridge/run-idle-test.sh
@@ -0,0 +1,18 @@
+#!/bin/sh
+
+cargo build
+
+# We've seen some segfauls from deno that require the interpreter running
+# for a period. The test here is to watch the bridge process and make sure
+# it is still alive after some time has passed.
+RUST_LOG="notset" target/debug/svix-bridge --cfg '{}' &
+SVIX_BRIDGE_PID=$!
+echo "Monitoring PID=$SVIX_BRIDGE_PID"
+sleep 15
+ps $SVIX_BRIDGE_PID > /dev/null
+if [ "$?" -ne 0 ]; then
+    echo "fail: process terminated prematurely"
+    exit 1
+fi
+echo "success: process stayed up"
+kill $SVIX_BRIDGE_PID