fluent
out_influxdb: documentation for tag prefix stripping
Documentation to go with https://github.com/fluent/fluent-bit/pull/9427
This removes a defined prefix from measurement names which might otherwise be shared between many measurements in the same data bucket. Compatibility is not affected, default behavior (with documented default strip_prefix "") is to strip zero character from the front which identical to the existing mechanism.
When writing to a range of different buckets, routing to the corresponding out_influxdb instances happens on tag matches. This change allows to match on tag prefixes, but strip them from the measurement name. This avoids having identical prefixes for all measurement names in the same data bucket.
To achieve this, read from char tag[] with an offset when writing the measurement name, provided the prefix matches the tag completely and the overlap is at most tag_length - 1 characters.
@fluent/chronosphere-tech-writers should we be ensuring we add YAML config as well now always?
@esmerel can you review please, but leave open as code PR https://github.com/fluent/fluent-bit/pull/9427 must merge first and is still open.
@ueli-g can you resolve the conflicts and provide YAML config example (if you need help with that, happy to provide)? I also nudged you on the code PR https://github.com/fluent/fluent-bit/pull/9427
I fixed the merge conflicts and made review updates while I was fixing it since I was already in there.
@eschabell thank you for the review. I added a yaml config example as requested and added more information on the behaviour. please let me know if there is anything else I can do to move this forward.
Config example needs updating to latest standard.
@patrick-stephens can you please have another look and let me know if this is more appropriate?