Manage a pool of connections about turbocharged.beanstalk HOT 5 CLOSED

hi, why consumer need to be isolated from producers?
beanstalkd not support that scenario or other reasons?

from turbocharged.beanstalk.

Hi @chrcao!

Beanstalk does support that scenario, but you have to be careful with it.

It has been a while since I worked with Beanstalk, but as I recall, Beanstalk responds to commands in the order they were sent. Let's say a client tried to be a consumer and producer at the same time, and sent these commands:

reserve
put 1 0 10 foobar

If there are no jobs in the tube for the reserve, then the put will be ignored until the reserve finally receives a job.

That's probably not a great outcome, and it would be easy to make that mistake. So, that's why I feel like a good client library should help the developer avoid accidentally causing that problem.

from turbocharged.beanstalk.

@jennings thanks for you answer, i got the point.
and another question when I review beanstalkd code. the data type of job id from beanstalkd is uint64.
in c#, the same capacity data type is ulong. Wil these cause problem when beanstalkd insert new job and return a job id more bigger than int(using in this client)?

from turbocharged.beanstalk.

Probably! 😄

Good catch. If you want to create a pull request for that, please do.

from turbocharged.beanstalk.

I no longer think this issue makes sense to do. The caller is perfectly capable of managing the number of connections they create.

But it does make sense to make BeanstalkConnection handle reconnection logic. Right now, if you create a worker and the connection dies, the worker dies. That isn't good.

from turbocharged.beanstalk.