Thoth Common

GitHub tag (latest by date) PyPI - Module Version PyPI - License PyPI - Downloads

A library used in project Thoth. It’s aim is to provide core utilities for logger setup, manipulation with datetimes and similar handy helpers. Namely, you can find:

  • operations used for cluster object management (e.g. obtaining logs of pods)

  • operations used for scheduling workload using Argo Workflows

  • Thoth’s configuration abstraction used for serialization/deserialization of Thoth’s config file across all the components

  • application logging setup

  • manipulation with date and time

Installation

This project is released on PyPI, so the latest release can be installed via pip or Pipenv as shown below:

pipenv install thoth-common

This library will automatically discover installed packages and enable Sentry integrations if you use Flask, SQLAlchemy or AIOHTTP. An exception is for Flask applications, that need to explicitly install sentry-sdk[flask] due to integrations dependencies.

Logging setup

To setup a logger in any of Thoth’s component (component that are namespaced with thoth), you can simply set an environment variable. The name of environment variable is constructed from module name. Let’s say you want to debug thoth.adviser.pipeline module, in that case you can set environment variable THOTH_LOG_ADVISER_PIPELINE=DEBUG which will cause loggers thoth.adviser.pipeline to be set to DEBUG mode. See log-levels documentation for more info. If a module has underscore in its name, replace it with double underscore in the environment variable name.

To setup a logger that is not introduced by a Thoth’s component, you can set THOTH_ADJUST_LOGGING environment variable. The format of this environment variable THOTH_ADJUST_LOGGING is a comma separated list where each entry is made out of a logger name and a corresponding log-level (”DEBUG”, “INFO”, “WARNING”, “ERROR” as for standard Python’s logging). These two are delimited by a colon, an example:

THOTH_ADJUST_LOGGING="flask:WARNING,alembic.migrations:ERROR"

Structured logging

The library will automatically detect when it is running inside an OpenShift cluster (based on STI_SCRIPTS_PATH environment variable that is inserted into the container if build the container was built in an OpenShift cluster), In such case, the library will setup structured logging suitable for automated logs aggregation (e.g. automated logs aggregation using the ELK stack). This behavior can be suppressed by setting environment variable THOTH_LOGGING_NO_JSON=1.

Ignoring reports from a logger

In some cases it’s expected to turn off reporting of some logger to Sentry. You can provide THOTH_SENTRY_IGNORE_LOGGER environment variable which holds a comma separated list of loggers that should be ignored when reporting errors to Sentry:

THOTH_SENTRY_IGNORE_LOGGER="thoth.adviser.resolver,thoth.adviser.run"

This is helpful if you want to report errors to users but not to Thoth application itself.

If you want some certain classes of error not to be reported to Sentry, you can can provide THOTH_SENTRY_IGNORE_EXCEPTION environment variable which holds a comma-separated list of exception classes to be ignored when reporting to Sentry:

THOTH_SENTRY_IGNORE_EXCEPTION="connexion.decorators.validation,builtins.ValueError"

Remember all builtin exception classes need to be specified as in the same manner as ValueError is specified above.

Crossroad

This documentation corresponds to implementation in version 0.36.4, documentation was generated on Oct 17, 2022.