flux-scribe¶
Warning
flux-scribe is written but still needs testing.
flux-scribe runs on rank 0 of the batch job and records every job event of the instance to a database, using the Flux journal consumer. Install the extra dependencies first:
flux python -m pip install --user "flux-batch[scribe]"
Then add the module and choose a database:
export FLUX_SCRIBE_DATABASE=sqlite:///flux-batch-job.db
flux python ./examples/flux_scribe_module.py
spec.add_module("flux-scribe")
Settings¶
Variable |
Default |
Description |
|---|---|---|
|
|
Database URL to write events to. |
|
unset |
Set to anything except |
|
unset |
Identifier stored with every recorded event. |