Open nuclearcat opened 1 month ago
Currently, we only support test results as enum values(pass/fail/miss/etc). We can enhance our node model to support additional numerical values, providing more detailed test results.
Identify and define the additional fields needed to store numerical values in the node model. @JenySadadia proposed to use "data" subfield.
These may be two different things. If what we want to support is tests whose result is a number instead of a discrete value (for instance, performance tests), then what needs to be reworked is the result field. If what we want is just to provide an additional field for a number besides the result, then yes, adding a data
field is a no-brainer.
Definitely it will be just additional field.
Many of the tests we run yield numerical values as output. For example, fluster tests may produce a "score" value, and boot time tests may return values in seconds or microseconds. Currently, we only support test results as enum values(pass/fail/miss/etc). We can enhance our node model to support additional numerical values, providing more detailed test results.
For reference, see the discussion: GitHub Issue #546.
Objective: The goal of this task is to extend the node model to support numerical values. The task can be divided into the following steps:
Design necessary field(s) in the node model
Add support to LAVA parser (optional)
kernelci-core/kernelci/runtime/lava.py
to handle the new numerical fields, if necessary. Probably @pawiecz can implement, or assist as LAVA expert.Implement data retrieval from LAB results
lava_callback.py
in thekernelci-pipeline
to retrieve and process numerical data from LAB results.Review and discuss KCIDB schema
Assist/submit PR for KCIDB changes (optional)
Implement data submission to KCIDB
KCIDB related changes is more expertise of @JenySadadia i believe.
Steps:
kernelci-core/kernelci/runtime/lava.py
.lava_callback.py
inkernelci-pipeline
.