Good to Know

Good to Know#

Updating workunits with many associated resources#

Note

Summary: Updating workunits and its associated resources

  1. Set workunit state to "processing".

  2. Update workunit and associated resources, as much as you need to.

  3. Set workunit state to "available".

When updating large workunits, it’s crucial to prevent state recomputation in B-Fabric by setting the workunit state to "processing" before making any changes. Failing to do so can lead to significant performance issues and complications, especially during iterative updates.

After all entities have been saved, you can set the workunit state to "available". However, be aware that B-Fabric will still perform a state recomputation at this stage, which might result in a different inferred state, such as "failed". The "processing" state is the only one that ensures state recomputation is bypassed during the saving process.

Logging level#

BfabricPy uses loguru for logging, since it makes logging very simple and informative. While it is recommended by its developers to make it opt-in, our logging is opt-out, but this can easily be adjusted as needed.

In your code#

You can configure loguru in your code:

import sys
from loguru import logger

# Configure a logger with custom settings (requires first removing the default logger)
logger.remove()
logger.add(sys.stderr, filter="bfabric", level="WARNING", colorize=False)

# Alternative: completely deactivate logs for bfabricPy
logger.disable("bfabric")

Check the documentation for logger.add for more options. By default, the logs go to standard error, which can be useful in some scenarios.

Alternatively, you can set the environment variable LOGURU_LEVEL, e.g. LOGURU_LEVEL=WARNING. However, this will affect all loguru loggers in your code and may thus not be the preferable approach.

Error Handling#

For comprehensive information about error handling, exception types, and patterns, see Error Handling.

In bfabricPy scripts#

Generally, to achieve a consistent output in our scripts, we initialize the logger by the following function:

from bfabric.utils.cli_integration import setup_script_logging

setup_script_logging()

This removes time stamps and line numbers, unless BFABRICPY_LOG_LEVEL=DEBUG is set.

The logging level can be controlled via the BFABRICPY_LOG_LEVEL environment variable:

Value

Effect

DEBUG

Show all log messages including debug output (with full loguru format)

INFO

Show INFO and above (default)

WARNING

Show WARNING and above

ERROR

Show ERROR and above

CRITICAL

Show CRITICAL only

OFF or 0

Disable all logging completely

Invalid values fall back silently to INFO.

Examples:

# Silence all log output (useful for scripting or piping)
BFABRICPY_LOG_LEVEL=OFF bfabric-cli api read resource

# Show only warnings and errors
BFABRICPY_LOG_LEVEL=WARNING bfabric-cli api read resource

# Enable full debug output
BFABRICPY_LOG_LEVEL=DEBUG bfabric-cli api read resource