From description to prescription: Unraveling log severity adjustments in open-source software
Bibliographic record
Abstract
Logs are vital to understanding a software system’s behavior, often being the only evidence available to investigate failures. Selecting a Log Severity Level (LSL) can be challenging for the following reasons: (i) the absence of knowledge about how logs are used in production, (ii) the lack of understanding of how critical an event is, and (iii) the lack of practical guidelines. This leads to frequent LSL adjustments during software development and evolution. Our goal is to investigate the LSL adjustments between system releases and explore methods to improve LSL classification. We analyzed the log statements from different releases of open-source systems, focusing on their LSL adjustments and examining the commit comments to understand the reasons for the adjustments. Our results show that most adjustments occur at the intersection of development and production environment logs. Furthermore, the main guiding factors for the adjustments are the experience and logging theory. Our contributions are (i) a description of trends and patterns in LSL adjustments and (ii) a set of 24 heuristics to guide the choice, review, and adjustments of LSL. We advise developers to adhere to the LSL purposes, routinely review LSL settings, and remain adaptable to their mutability. • The severity level of log statements can change as the software evolves. • Severity level adjustments occurring between system releases tend to be more experience-oriented rather than based on logging theories. • Avoiding the overproduction of log data is one of the main reasons for adjusting severity levels in the systems investigated. • There is a tendency towards one-degree adjustments with an emphasis on adjustments between the Debug and Info levels. • From our research, we have derived a set of 24 heuristics designed to guide the choice, review, and adjustment of log severity levels.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.020 | 0.176 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.008 | 0.006 |
| Science and technology studies | 0.002 | 0.003 |
| Scholarly communication | 0.007 | 0.016 |
| Open science | 0.002 | 0.005 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".