[Messenger] Fix some syntax issue

[javiereguiluz]

In addition to the syntax issue, there's another pending issue: which option does it refer to? I can't see the keepalive option in the list of options above it.

So, is the option missing ... or should we delete this versionadded directive?

[mdoutreluingne]

I think this is an oversight, because for Amazon SQS transport, the keepalive option still exists.

I've also noticed that the option keepalive is missing from the list of options for Beanstalkd Transport

[javiereguiluz]

Maxime, thanks for reviewing this. I've added some content based on the docs of these services. But, I have no experience in this, so please review it. Thanks!

[mdoutreluingne]

I have little experience in this and I'm not familiar with this transports, especially as there can be differences depending on the type of transport.

In my opinion and according to my research, it looks good but I'm not sure.

Need review for the keepalive option for the Amazon SQS transport and Beanstalkd transport, can you review this please ?

cc @HypeMC @valtzu

[valtzu]

I don't think we added anything to the transport options – what was added was support for --keepalive in messenger:consume command, for these specific transports.

How SQS works with --keepalive:

bin/console messenger:consume --keepalive=10 async

framework:
  messenger:
    transports:
      async: 'sqs://...?visibility_timeout=15'

Every 10 seconds, for all the messages that are still being processed, the visibility of a message is re-set to visibility_timeout.

When using --keepalive, visibility_timeout must be bigger than --keepalive value.

This mechanism allows to keep visibility_timeout rather small, thus, in case the worker dies, the message is redelivered faster. Without --keepalive, a message that takes more than visibility_timeout to process would lead into double-processing as SQS would redeliver it while it's already being processed.

[HypeMC]

The default keepalive interval is actually 5 seconds. When the option is used without an explicit value, it activates the keepalive mechanism with that default value. You can optionally set a different interval by providing a value for the option.

[HypeMC]

I don't think we added anything to the transport options – what was added was support for --keepalive in messenger:consume command, for these specific transports.

Ok, I just noticed that this was added to the list of transport options. @valtzu is right, there were no options added to the transports themselves, only to the messenger:consume and messenger:failed:retry commands. These console command options tell the Worker to asynchronously use the transport's keepalive mechanism if it's implemented.

[javiereguiluz]

Thanks a lot folks! I now understand this. I made some changes based on your comments.

[mdoutreluingne]

Thanks a lot for explaining this, I understand better.