Install pm2 for better handling of uncaught runtime errors (#24)

Dockerfile

@@ -96,8 +96,11 @@ RUN chmod -R +x /docker-entrypoint.d/start-node-prod.sh
 # Set the environment to production for install & runtime.
 ENV NODE_ENV=production
 
+# Install pm2 for better handling of uncaught runtime errors. 
+RUN npm install pm2 -g
+
 # Install the node modules.
-RUN npm ci --only=prod 
+RUN npm ci --only=prod
 
 # Remove the npm package manager.
 RUN apt remove -y curl npm unzip

conf/entrypoint/start-node-prod.sh

@@ -1,4 +1,4 @@
 #!/bin/bash
 
-nohup node /usr/local/bin/node/server.js &
-
+# pm2 will restart the node process on unhandled errors, `--no-daemon` will attach to application log
+pm2 start /usr/local/bin/node/server.js --no-daemon &

conf/node/constants.js

@@ -53,6 +53,7 @@ export const allowedTargetAgencies =
 
 export const sensitiveFiles = [
   "cookies.txt", // Contains the JWT and CloudFront cookies.
+  "hts-in_progress.lock", // Contains the JWT.
   "hts-log.txt", // Has the httrack command line arguments - this includes the JWT.
   `hts-cache/doit.log`, // Has the httrack command line arguments - this includes the JWT.
   `hts-cache/new.zip`,

conf/node/controllers/main.js

@@ -30,11 +30,16 @@ export const main = async ({ url, agency, depth }) => {
 
   runHttrack(httrackArgs);
 
-  await waitForHttrackComplete(paths.fs);
+  const { timedOut } = await waitForHttrackComplete(paths.fs);
+
+  if (timedOut) {
+    console.error("Httrack timed out", { url: url.href, agency, depth });
+    return;
+  }
 
   // Remove sensitive files - before syncing to S3
   await Promise.all(
-    sensitiveFiles.map(file => fs.rm(`${paths.fs}/${file}`, { force: true }))
+    sensitiveFiles.map((file) => fs.rm(`${paths.fs}/${file}`, { force: true })),
   );
 
   // Sync the snapshot to S3