Autosubmit databases


Autosubmit stores information about its experiments and workflows in SQLite databases and as serialized Python objects (pickle files). These are distributed through the local filesystem, where Autosubmit is installed and runs.

There is one central database that supports the core functionality of experiments in Autosubmit. There are other auxiliary databases consumed by Autosubmit and the Autosubmit API, that store finer-grained experiment information.

The name and location of the central database are defined in the .autosubmitrc configuration file while the other auxiliary databases have a predefined name. There are also log files with important information about experiment execution and some other relevant information such as experiment job statuses, timestamps, error messages among other things inside these files.


The <EXPID> is an experiment ID. The location of the databases of other files can be customized in the .autosubmitrc configuration file.

High level view of the Autosubmit storage system

Core databases


Default location




The main database of Autosubmit. The location can be customized in the autosubmitrc file.



Deprecated API. Used by Autosubmit API with Autosubmit 3.x. Kept for backward compatibility for now.

Auxiliary databases

These databases complement the databases previously described for different purposes. Some of them are centralized in the $AS_METADATA directory (defined in the .autosubmitrc config file) while others are present inside each experiment folder.

Databases in the $AS_METADATA directory


Default location




Used by the GUI to improve the graph visualization. Populated by an API worker.



Used by the GUI to display edge lists. Populated by an API worker.



Stores the status of the partition where Autosubmit databases and experiment files are stored. Populated by an API worker.



Stores experiment metrics and historical information. Populated by Autosubmit.

Databases in each experiment directory


Default location




Stores information about the wrappers configured in the experiment. Empty if no wrappers configured.



Deprecated. Used in Autosubmit 3.x, now replaced by the database used by the Autosubmit API (described above).

Other files

Autosubmit stores Pickle files (e.g. $HOME/autosubmit/<EXPID>/pkl/job_list_<EXPID>.pkl) with the job list of experiments. In the event of a crash, or if the user stops the experiment, that Pickle file is used in order to be able to restore the experiment to its latest status.

There are also update list files, used to change the status of experiment jobs without stopping Autosubmit. These files are plain text files, and also present in the experiment directory.