Ryan Clark

My expectation was that if one task fails, then the queue would stop processing additional tasks. I'm not quite sure whether or not this is in the same spirit as...

Do you have any way to observe the S3 connections or S3 errors more directly? All the error codes you've got here appear to be from downstream applications that are...

I think I'd have to take it to AWS support. You might try to check for `x-amz` headers in the HTTP response to see if S3 is trying to tell...

The result here would actually be the same as `exit 0`: remove the message from the queue, don't consider the job a failure. Any reason you can't just use `0`...

When I am my best self, I say "job" to indicate an end-to-end process that watchbot performs, and I say "part" when I'm talking about a job that's been split...

oooh that heartbeat solution is really a good one. The watcher can do that for sure. We would want to change the watchbot options a little though. You'd give the...

Note: I think that there is a limit to how long you can defer visibility. 12 hours as I recall.

If you have an option called `jobDuration`, but that does not represent any cap on a worker's duration, what is the purpose of the option? With the hearbeat solution I...

> Set the visibility timeout to be this maxJobDuration value. I would suggest we pull from the queue with a 3 min visibility timeout, always. Then, every 2 minutes the...

The most straightforward option to consider is to make 12 hour a hard job duration limit. If it gets to 12 hours, the child process gets killed and the job...