Beanstalk in a centralized server, how to avoid duplicate work for a worker?

Question

I have a server that runs Beanstalk, and some independent servers that runs the workers, in PHP, with Pheanstalk.

From the moment a worker A get a job :

$job = $pheanstalk->watch('tube')
    ->ignore('default')
    ->reserve();

$data = json_decode($job->getData(), true);

And the moment it delete the job ($pheanstalk->delete($job);), it could happens a few ten seconds.

Does Beanstalk know the job is being processed and no other worker will have it, or will I have a concurrency problem? (two worker taking the same job).

Thank you for your help.

score 2 · Accepted Answer · edited Jan 11 '16 at 17:54

2

BeanstalkD knows the job that is being processed, so it won't serve the job to another worker while it is processing it.

edited Jan 11 '16 at 17:54

Nathaniel Ford

20,545
20
91
102

answered Jan 11 '16 at 17:33

Yaniv Davodovich

36
3

Your answer is exactly what I want to hear, but do you have link to confirm your answer? – Cyril N. Jan 12 '16 at 07:30
1

When a job has been reserved it's status changed to _reserved_. "reserve" command serve only jobs with _ready_ status. https://github.com/kr/beanstalkd/blob/master/doc/protocol.txt – Yaniv Davodovich Jan 12 '16 at 09:13
Perfect! Thank you :) – Cyril N. Jan 12 '16 at 10:08

Beanstalk in a centralized server, how to avoid duplicate work for a worker?

1 Answers1