WhyLabs: whylogs rolling logger wrote empty profiles for segmented schemas

WhyLabs: whylogs rolling logger wrote empty profiles for segmented schemas

TL;DR

If your whylogs profiles read back empty, check whether you are mixing the rolling logger with segmented schemas on an old version, and upgrade past v1.1.13. When the writer and the reader disagree on the profile format, you get silent empty reads, not errors. Context: From GitHub issue whylabs/whylogs#965, fixed in v1.1.13 (PR #971).

Use this when you run into the situation in the title.

When not to use this skill: unrelated tasks. It covers only the procedure above.

Compatibility

The steps above apply to the commands named in them. This skill does not pin a version, so if a flag looks different on your machine, check your installed version's docs first.

Details

Using the rolling logger with a DatasetSchema that has segments produced completely empty profiles when read back. The root cause: segmented profiles were serialized in the old v0 format while why.read expected the v1 format, so every file came back empty even though logging appeared to work. The one-shot why.log call worked fine, which made it look like a logger problem rather than a serialization mismatch.

Source: From GitHub issue whylabs/whylogs#965, fixed in v1.1.13 (PR #971). Using the rolling logger with a DatasetSchema that has segments produced completely empty profiles when read back. The root cause: segmented profiles were serialized in the old v0 format while why.read expected the v1 format, so every file came back empty even though logging appeared to work. The one-shot why.log call worked fine, which made it look like a logger problem rather than a serialization mismatch.

Verified resolution: If your whylogs profiles read back empty, check whether you are mixing the rolling logger with segmented schemas on an old version, and upgrade past v1.1.13. When the writer and the reader disagree on the profile format, you get silent empty reads, not errors.

Source URL: https://github.com/whylabs/whylogs/issues/965