Each workflow maintains its own log of time-stamped events in the
workflow log directory (
The information logged here includes:
Event timestamps, at the start of each line
Workflow server host, port and process ID
Workflow initial and final cycle points
Workflow start type (i.e. cold start, warn start, restart)
Task events (task started, succeeded, failed, etc.)
Workflow stalled warnings.
Client commands (e.g.
Information relating to the remote file installation is contained in a separate log file, which can be found in
Workflow log files are primarily intended for human eyes. If you need to have an external system to monitor workflow events automatically, use:
The GraphQL interface (can be accessed via GraphiQL in the Cylc GUI).
The sqlite workflow run database (see Workflow Run Databases)
Rather than parse the log files.
The Scheduler Log also records messages sent by
cylc message allowing you
to add custom messages to this log.
For example, if your task contained the following code:
cylc message -- "ERROR:some_function failed."
Your log should produce output similar to:
ERROR - [21491012T0410Z/mytask running job:01 flows:1] (received)some_function failed ERROR at 2023-04-14T11:36:35+01:00
Severity levels are the same as those used by Python’s logger:
Messages logged at “DEBUG” will only appear in the scheduler log if the
workflow is played with
cylc play --debug.