Error: ENOENT: no such file or directory, unlink

[jangrewe]

Describe the bug
We're running smtp-relay in Docker on an EC2 instance, and the container regularly crash with this error (but varying filenames, of course):

Sep 12 07:17:10 gitlab docker[1603915]: node:fs:1883
Sep 12 07:17:10 gitlab docker[1603915]:   handleErrorFromBinding(ctx);
Sep 12 07:17:10 gitlab docker[1603915]:   ^
Sep 12 07:17:10 gitlab docker[1603915]: Error: ENOENT: no such file or directory, unlink '.email_20230912T071709979Z.eml'
Sep 12 07:17:10 gitlab docker[1603915]:     at Module.unlinkSync (node:fs:1883:3)
Sep 12 07:17:10 gitlab docker[1603915]:     at PassThrough.<anonymous> (file:///smtp-relay/lib/server.js:188:20)
Sep 12 07:17:10 gitlab docker[1603915]:     at process.processTicksAndRejections (node:internal/process/task_queues:95:5) {
Sep 12 07:17:10 gitlab docker[1603915]:   errno: -2,
Sep 12 07:17:10 gitlab docker[1603915]:   syscall: 'unlink',
Sep 12 07:17:10 gitlab docker[1603915]:   code: 'ENOENT',
Sep 12 07:17:10 gitlab docker[1603915]:   path: '.email_20230912T071709979Z.eml'
Sep 12 07:17:10 gitlab docker[1603915]: }
Sep 12 07:17:10 gitlab docker[1603915]: Node.js v18.17.1

The majority of mail it accepts gets handled without any issues, but this happens every once in a while (e.g. twice in 12 hours now).

To Reproduce
Steps to reproduce the behavior:

1. run smtp-relay
2. send mails (in our case from GitLab, but i don't think this matters)
3. See error

Expected behavior
Emails get relayed and the container keeps running.

[loopingz]

The only unlink used in the server code is here:

    // @ts-ignore
    stream.on("end", async () => {
      await this.filter("Data", session, [session]);
      session.email = await simpleParser(fs.createReadStream(session.emailPath));
      await this.onDataRead(session);
      if (!this.config.keepCache) {
        fs.unlinkSync(session.emailPath);
      }
      callback();
    });

I'm still unsure why it cannot delete it, I will update code to have a try/catch around the unlink today, in the mean time you can try to add a keepCache: true to your configuration

[jangrewe]

Thanks, i've added this to the config and will keep an eye on the log and any updates. 😃

[loopingz]

The 1.3.3 version should not have the issue, please keep me update, and if you like the project please star it. It helps with visibility

[jangrewe]

Thank you! I've just deployed the updated image and now i'm waiting for the first error (without crash), but i'm confident that your changes will be doing what they're supposed to. 😀