unable to report errors more than once when handlers wrapped with "withSentry"

[demsey2]

At this stage, I am not sure if the issue is with this library or @sentry/node
so I have a lambda with the following code

import withSentry from "serverless-sentry-lib";
export const handler = withSentry(async (event: S3Event, context: Context) => {
  throw new Error('test');
});

so if I call that lambda (no instance/cold start), it will report the error, but when I try again and the same lambda instance is called then the error is not reported. I thought it might be because is the same message but I tried

throw new Error(`test with dynamic error ${Date.now()}`);

and no luck with that

then I just added Sentry.init

export const handler = withSentry(async (event: S3Event, context: Context) => {
  Sentry.init({ dsn: process.env.SENTRY_DSN });
  throw new Error('test');
});

and this is reporting errors as many times as I call that lambda

my dependencies

"serverless-sentry": "^2.0.2",
"serverless-sentry-lib": "^2.0.1",
"@sentry/node": "^5.18.1",

when I try to call

let client = Sentry.getCurrentHub().getClient();
console.log(client.getOptions());

I got

{
  dsn: '...',
  release: '984ec86',
  environment: 'dev',
  defaultIntegrations: [
    InboundFilters { _options: {}, name: 'InboundFilters' },
    FunctionToString { name: 'FunctionToString' },
    Console { name: 'Console' },
    Http { name: 'Http', _breadcrumbs: true, _tracing: false },
    OnUncaughtException {
      _options: {},
      name: 'OnUncaughtException',
      handler: [Function]
    },
    OnUnhandledRejection {
      _options: [Object],
      name: 'OnUnhandledRejection'
    },
    LinkedErrors { name: 'LinkedErrors', _key: 'cause', _limit: 5 }
  ],
  enabled: false
}

that flag enabled: false doesn't exist when I call lambda first time

[arabold]

Interesting bug report. Thanks! This will certainly need some additional investigation.

[arabold]

I've just published 2.1.0 that should address this issue. I'm not perfectly sure if this is actually a Sentry issue, but when the Lambda finishes we now call flush instead of close to keep the Sentry instance intact. This should fix the problem.