Extracting PM metrics on Kafka Cluster results in duplicate files due to different cases in the same metric names.

[secureaniruddha]

One of our customers is trying to extract PM metrics logs and has encountered duplicate files due to different cases with the same metrics names. This means one file contains metrics entries in uppercase and another in lowercase. They want to know if all java_lang related PM metrics entries can be configured in lowercase to avoid file duplication.

Is it possible to configure all java_lang related metrics entries in lowercase, and is it recommended?

metrics. zip

[dhoard]

@secureaniruddha you can use lowercaseOutputName: true to make all metric names lowercase, but there is no per-metric configuration.

[secureaniruddha]

Hi @dhoard Thanks for the swift reply. If there is no option for per metric configuration, is there any way which ensure there should be a single file in any case to avoid duplication?

[dhoard]

@secureaniruddha there is no per-metric configuration to convert the metric name to lowercase. Typically lowercaseOutputName: true is used for metric name consistency.

Can you elaborate on what you mean by "file duplication"?

[secureaniruddha]

@dhoard After upgrading files can be generated ok, but extracting pm metrics files has duplicated files in windows. The duplicated files' content are different, and file names differences are: capital letters and lowercase.
On linux this leads to an inconsistency because different case filenames are treated as different files (it still is an issue of different contents and writing to different places). On Windows it's more pressing issue, since same name different case files overwrite one another, and the metrics data is essentially lost.

[dhoard]

@secureaniruddha This sounds like an issue with the CCRC tool, since the exporter only exposed metrics via HTTP(S).