Monitoring EaaSI Systems
The EaaSI platform does not currently provide much in the way of advanced methods for monitoring system health or usage. The following page contains current recommendations for methods or work-arounds for EaaSI system administrators to gain some basic insight in regard to their EaaSI installation.
System Health
OpenSLX is actively investigating the use of Microsoft Playwright to allow for headless testing of particular Emulation-as-a-Service API endpoints and services; these tests would be broadly applicable to EaaSI deployments as well. In the meantime, administrators interested in active or automated monitoring of the health of their EaaSI installation can at least use HTTP GET requests to poll a handful of API endpoints; a 200 OK request from any of the following URLs indicate that the relevant EaaSI stack component is nominally up and reachable:
https://<eaasi-domain>/emil/admin/buildInfocan be used to poll the EaaS Java server
https://<eaasi-domain>/auth/realms/master/.well-known/openid-configurationcan be used to poll the Keycloak user management component
https://<eaasi-domain>/dashboardcan be used to poll the EaaSI UI
https://<eaasi-domain>/admincan be used to poll the Demo UI (if relevant)
Usage Metrics
Sessions Metadata
The Emulation-as-a-Service application logs minimal, anonymized metadata on emulation sessions to allow system administrators some insight into which Environment resources have been run on their deployment, how often, and for how long.
This metadata is logged to a CSV file located at <eaasi-home-dir>/server-data/sessions.csv. Only system administrators/users with direct (e.g. SSH) access to the server EaaSI is deployed on can access this file; it is not accessible via a public-facing interface (EaaSI UI, Demo UI, or API).
This file logs the following metadata points for every emulation session ever run on the server (unlike the request stats on the Demo UI dashboard, this session data persists between restarts of the eaas service, so represents a complete picture of the server’s usage):
Timestamp (UTC) for the start of the session
Timestamp (UTC) for the end of the session
UUID for the user account that requested/started the session
UUID for the requested Environment resource
UUID for any Content/object resource associated with the requested Environment
System administrators can map UUIDs for Environment and Content resources to human-readable resource names and other helpful/identifying metadata via the EaaSI API.
Using stats-sessions.py
The OpenSLX team has also created a simple Python script to assist with quickly parsing the metadata logged in sessions.csv to a couple of commonly-requested datapoints.
$ stats-session.py [--filename sessions.csv] [sessions/max-sessions]
By default, stats-session.py will assume the user is running the script on their EaaSI server and run on the file location /eaas*/server-data/sessions.csv. Use the --filename directive as shown above to specify a different file location/path.
Users can then specify one of two reports to generate from the sessions.csv.
The sessions action repeats much of the same metadata contained in the file but presented in a more human-readable form, e.g.
2021-05-24 12:55:46.025000+00:00 097ccc71-db3d-46d8-b006-8f04766b7100 117.713
Where these values represent:
The UTC timestamp for when the emulation session was started, in ISO 8601 format for legibility
The UUID for the environment run in the session
The duration of the session, in seconds
The max-sessions action quickly calculates the maximum number of simultaneous sessions for every day that the EaaSI server has been active, e.g.:
2021-05-25 12:55:46.025000+00:00 4
2021-05-26 19:12:48.167000+00:00 1
2021-05-27 01:37:35.529000+00:00 0
2021-05-28 04:28:34.209000+00:00 7
This information is primarily useful for resource allocation for the EaaSI deployment (i.e. calculating whether more CPU/RAM should be allocated to the server given any patterns of heavy simultaneous usage).
By default, stats-session.py will simply write these reports as tab-separated values to standard output. If you would like to save the reports, redirect the output to file, e.g.
$ stats-sessions.py max-sessions > max-sessions.tsv