[Understanding CLP IR] How are Fluenbit generated IRs turned to Archives

[shadowshot-x]

Hi CLP Team.

I have been exploring CLP and trying to understand more about the technology. I am able to generate zstd compressed IRs and upload to S3.

The piece which feels missing is how to convert these IRs to archives so I can Query them using Presto.

I am using CLP 0.7.0 and latest Fluentbit clp plugin

Command: ./clp-s c /mnt/logs/archives/ --timestamp-key '@time' /mnt/logs/test_0_2026-01-08T11_40_27+05_30_139bce15-095c-4ea3-b8ce-8a11ecf77aa6.ir.zst
Error

2026-01-08T16:54:54.040+00:00 [error] Encountered curl error while ingesting /mnt/logs/test_0_2026-01-08T11_40_27+05_30_139bce15-095c-4ea3-b8ce-8a11ecf77aa6.ir.zst - Code: 3 - Message: 
2026-01-08T16:54:54.041+00:00 [error] Could not deduce content type for input /mnt/logs/test_0_2026-01-08T11_40_27+05_30_139bce15-095c-4ea3-b8ce-8a11ecf77aa6.ir.zst
2026-01-08T16:54:54.050+00:00 [error] Encountered error while parsing input.

Also tried decompressing the ZST
Command: ./clp-s c /mnt/logs/archives/ --timestamp-key '@time' /mnt/logs/test_0_2026-01-08T11_40_27+05_30_139bce15-095c-4ea3-b8ce-8a11ecf77aa6.ir
Error

2026-01-08T16:56:55.684+00:00 [error] Encountered error while creating kv-ir deserializer: (93) - Protocol not supported
2026-01-08T16:56:55.694+00:00 [error] Encountered error while parsing input.

If we can convert fluentbit generated IRs to archives, that will be a huge win for the platform!