#create_index! produces error output under correct operation

[dogweather]

Given that there's no index, when this is executed:

klass.__elasticsearch__.create_index! force: true

then this is sent to stderr:

[!!!] Index does not exist (Elasticsearch::Transport::Transport::Errors::NotFound)

I believe that nothing should be sent to stderr in this particular case, because we've asked the lib to "create an index, over-writing any potentially existing one". And so there's no error when none exists.

I'm willing to contribute a PR if this issue is approved.

(Relevant code)

[andyw8]

@dogweather any luck with solving this? I'm not super familiar with with ES but this message pollutes my test output.

[karmi]

Hi @andyw8, I think we've debated adding some "quiet mode" in some issue, but I don't recall the reference now.

Looking at the code now, it comes from delete_index!, where we should use the ignore option in the Elasticsearch client, instead of capturing the exception like this.

[webhive]

Any progress on it? Very annoying!

[Kani999]

Still not solved?

[shawndeprey]

Still running into this issue as well. The quiet mode that @karmi mentioned above would be very useful.

[estolfo]

@shawndeprey @Kani999 @andyw8 @dogweather how does this solution look?
I use the client transport logger to log the message. That way, the user has control over whether it's printed via the log level.

[shawndeprey]

@estolfo If I understand what client.transport is this seems reasonable to me as this will follow the logging set in our application configuration. 👍

[estolfo]

@shawndeprey client.transport is the client transport layer, which is where the logger would be set if you set one on an Elasticsearch::Client. That said, it sounds like we're on the same page = )

[estolfo]

Closing as this has been resolved in the linked PR.