[FeatureRequest] Make number of retries configurable for /nodes and potentially other relevant http calls

[dwco-z]

Is your feature request related to a problem? Please describe.
Currently rqlite uses defaultMaxRetries(8) during GetNodeAPIAddr on the cluster client. Considering that unreachable nodes like shutdown rqlite instances will keep failing during the first + 8 retry attempts, having the option to configure this number of retries and possibly decrease it might be useful to avoid having too many retries being ineffectively executed.

Describe the solution you'd like
A possible solution could be making the defaultMaxRetries configurable

[otoolep]

I may just start by setting the default retry value to 0. This is the second issue that's been filed about the retry behavior, so I think it's time to start removing default retries. It's always been an implementation detail, so I wouldn't consider it a breaking change.

[otoolep]

I do plan to allow is to be set on a request-by-request basis, but may just change the default to 0 to see if tests still pass.

[dwco-z]

I do plan to allow is to be set on a request-by-request basis, but may just change the default to 0 to see if tests still pass.

I may just start by setting the default retry value to 0. This is the second issue that's been filed about the retry behavior, so I think it's time to start removing default retries. It's always been an implementation detail, so I wouldn't consider it a breaking change.

Right, take makes sense and it would help with my issue

[otoolep]

@dwco-z -- this is now fixed on master. You will get zero retries without doing anything, but can add retries=N to your request (as a query parameter) where N is an integer.

[otoolep]

I can make a release, but perhaps you just want to build master to check it out.

[dwco-z]

Thanks @otoolep. Sure, I'll build master to test it

[dwco-z]

Everything looks good, now I don't have /nodes retries happening on the rqlite's side per default