Replica settings should not be set to zero

[jayhilden]

Currently, when the Elmah index is setup in ElasticSearch the following defaults are overridden:

Replicas = 0 Shards = 1 merge.policy.merge_factor = 10 --note: this is the Elastic Default search.slowlog.threshold.fetch.warn = 1s

https://github.com/elmahio/Elmah.Io.ElasticSearch/blob/06433749c0e75757a95b56c061ab946cf3f4c14e/Elmah.Io.ElasticSearch/ElasticSearchErrorLog.cs#L122

In a production environment where there is more than 1 node setting replicas = 0 means that a copy of the data would not exist on the other nodes in the cluster. This is not ideal.

ElasticSearch support recommends leaving all of these at the default values and only tweaking them if necessary in production.

[ThomasArdal]

You are absolutely right. The code should use Elasticsearch defaults. The only reason for not implemented this way is, that this code is written years ago and I haven't spend time updating it. This error logger is 100 % community driven at the moment, so feel free to send a pull request :smile:

[jayhilden]

And by community driven do you mean driven by me? :) I'll have a pull request for you later today.

[ThomasArdal]

You ARE the community in this case :p What I'm saying is, that I no longer contribute to this project. I've moved all of my personal stuff to elmah.io. I still see a need for a project like this, why it's great to have people like you contributing!

[jayhilden]

@ThomasArdal I'm more than happy to help! I'm converting our organization from Elmah over SQL to Elmah over ES so I'll have some good feedback about what is working and what can be improved. We love the idea of having centralized logging and we love Elastic!

[ThomasArdal]

Sounds great!