A guide for Guru Administrators

• It listens to the local Guru Server and turns each Guru user change into a LOGON and LOGOUT event.
• A Map table charges each user and host to a location, cost center, project and division. A shared login such as Engineer is still charged to the right site.
• Remote Users can use unique User IDs to identify their unique location, cost center, or project.
• Events are written as small JSON files and uploaded by FTP or FTPS to a landing folder, for example /incoming/cassini.
• Commercial 3rd Party Reporting Tools (like Microsoft Power BI) reads that folder as a Folder source.

1. Choose Apps > Guru Browser.
2. Choose Guru Import List and press Get Import File(s).
3. Choose the Cassini Monitor .gzp and press Import to select the file, then press Import again to start importing to Guru.
4. Check that the Cassini Monitor v# button now appears in Apps button.

1. Choose Apps > Cassini Monitor v#. The monitor window opens and shows the live log.
2. Choose Edit > Preferences.... The editor has three tabs: FTP, Monitor and Map. The monitor keeps running while it is open.
3. On the FTP tab, enter the host, mode (ftp or ftps), port, user, password and directory (for example /incoming/cassini). Typing a host turns on ftp.enabled.
4. Press Test FTP.... It connects, logs in, opens or creates the directory, and writes and deletes a test file. It reports "FTP connection OK" or names the setting to fix.
5. Set the upload interval with the slider: Immediate to Never, 20 minutes by default.
6. On the Map tab, add one row per user and host (see the example below). The first matching row wins. If the user credentials are known, from Guru, choose System > Logoff then proceed to Logon with all known users. This will pre-populate the table with default blank values that can be easily updated.
7. Press Save. This stores the settings in Guru as a RiPreferences object named RB0J6C1A, owned by this tester's Guru.
8. Choose Edit > Refresh settings to apply them now without a restart.

• To set up many testers, configure one, use Export Settings... to write a monitor-settings.properties file, and load it on the others with Import local files.... The exported file holds ftp.password in plain text, so store it carefully.
• Use ftps where the server supports it. Plain FTP sends the password unencrypted.
• Anyone who can read the RiPreferences object can read the FTP password. Use an upload-only FTP account.

1. In the monitor window, open the File menu.
2. Choose Enable Guru Autostart Agent.... This saves a RiAutoStart object (run mode startup, application RB0J6C1A) owned by this tester's Guru, titled Cassini Monitor autostart <host>.
3. Close Cassini Monitor and choose any setting.
4. Restart Guru by choosing System > Shutdown. Guru will automatically restart.
5. Open RiApps/GuruApps/RB0J6C1A/logs/cassini-monitor.log and look for Guru launch: autostart ..., then Started monitor ... and Monitoring silently in the background.

• The File menu shows Enable while no agent is active and Disable Guru Autostart Agent... while one is. Disable saves a new revision marked RiAutoStart_OFF; Enable turns it back on rather than creating a second agent.
• Choosing Apps > Cassini Monitor v# while the background monitor runs opens the window, which takes over the same session. The earlier log appears above the live lines.
• Closing the window asks Run in background, Stop monitoring or Cancel. Choose Run in background to keep recording. Stop monitoring ends the session, and nothing is recorded until Guru starts the app again.
• If Guru exits, the monitor keeps running and reconnects when Guru comes back.

Example Output JSON

Example Output JSON
Example Output JSON
Each upload is one UTF-8 file holding a JSON array of flat event objects, schema cassini-monitor/1. Files are named cassini_<host>_<yyyyMMddTHHmmssZ>_<seq>.json. Each file goes up as .part and is then renamed, so a reader never sees a half-written file.
A session for the shared Engineer account, one LOGON and its LOGOUT:
[
{
"schema": "cassini-monitor/1",
"eventId": "cassini-01-1790604191000-1",
"type": "LOGON",
"timestampUtc": "2026-09-28T14:03:11Z",
"epochMs": 1790604191000,
"host": "cassini-01",
"ip": "10.0.0.5",
"os": "Linux",
"osVersion": "2.6.18-419.el5",
"osUser": "engineer",
"guruUser": "Engineer",
"guruUserId": "63TX4N0",
"location": "Santa Clara",
"costCenter": "CC-4100",
"project": "Cassini-Dev",
"division": "Engineering",
"mapped": true,
"sessionId": "cassini-01-1790604191000",
"sessionStartUtc": "2026-09-28T14:03:11Z",
"durationSec": null,
"reason": null,
"agentVersion": "2"
},
{
"schema": "cassini-monitor/1",
"eventId": "cassini-01-1790607791000-2",
"type": "LOGOUT",
"timestampUtc": "2026-09-28T15:03:11Z",
"epochMs": 1790607791000,
"host": "cassini-01",
"ip": "10.0.0.5",
"os": "Linux",
"osVersion": "2.6.18-419.el5",
"osUser": "engineer",
"guruUser": "Engineer",
"guruUserId": "63TX4N0",
"location": "Santa Clara",
"costCenter": "CC-4100",
"project": "Cassini-Dev",
"division": "Engineering",
"mapped": true,
"sessionId": "cassini-01-1790604191000",
"sessionStartUtc": "2026-09-28T14:03:11Z",
"durationSec": 3600,
"reason": "userLogout",
"agentVersion": "2"
}
]
The agent also writes AGENT_START and AGENT_STOP events, with null user and session fields, which show when monitoring was running:
{"schema":"cassini-monitor/1","eventId":"cassini-01-1792501200000-1","type":"AGENT_START","timestampUtc":"2026-10-20T13:00:00Z","epochMs":1792501200000,"host":"cassini-01","ip":"10.0.0.5","os":"Linux","osVersion":"2.6.18-419.el5","osUser":"engineer","guruUser":null,"guruUserId":null,"location":"Unassigned","costCenter":"CC-0000","project":"Unassigned","division":"Unassigned","mapped":true,"sessionId":null,"sessionStartUtc":null,"durationSec":null,"reason":null,"agentVersion":"1"}
Key fields:

Field	Meaning
eventId	Unique per host; use it to remove duplicates
type	LOGON, LOGOUT, AGENT_START or AGENT_STOP
timestampUtc / epochMs	When the event happened, in UTC
host, ip	The tester
guruUser, guruUserId	Guru logon name as typed (compare ignoring case) and Guru UserID
location, costCenter, project, division	From the Map table, at the time of the event
mapped	false when no Map row matched
sessionId	Shared by a LOGON and its LOGOUT
sessionStartUtc	Start of the session, on LOGON and LOGOUT
durationSec	Session length in seconds, on LOGOUT only
reason	Why the session ended, on LOGOUT only
agentVersion	Cassini Monitor release number

LOGOUT reasons:

reason	Cause
userChanged	A different user logged on
userLogout	Guru reported no user
guruExit	Guru exited
connectionLost	The connection to Guru dropped
agentStop	The monitor was stopped normally
agentRestart	A session left open by a crash or power loss, closed at the last heartbeat

Map values are resolved when each event is written. Editing the Map table changes only later events; old files keep what was true at the time.

Example: Building dashboards and reports in Power BI

Example: Building dashboards and reports in Power BI
Point Power BI at the FTP landing folder, build one Events table, and treat every LOGOUT row as a finished session.

1. Load the events
In Power BI Desktop choose Get Data > Folder and point it at the landing folder, for example through an SMB share on the FTP host. Then replace the query with this in the Advanced Editor:
let
Source = Folder.Files("\\ftp-host\incoming\cassini"),
JsonOnly = Table.SelectRows(Source, each Text.EndsWith([Name], ".json")),
Parsed = Table.AddColumn(JsonOnly, "Rows", each Json.Document([Content])),
Rows = Table.ExpandListColumn(Table.SelectColumns(Parsed, {"Rows"}), "Rows"),
Events = Table.ExpandRecordColumn(Rows, "Rows",
{"eventId","type","timestampUtc","host","osUser","guruUser","guruUserId",
"location","costCenter","project","division","mapped",
"sessionId","sessionStartUtc","durationSec","reason","agentVersion"}),
Typed = Table.TransformColumnTypes(Events, {
{"timestampUtc", type datetimezone}, {"sessionStartUtc", type datetimezone},
{"durationSec", Int64.Type}, {"mapped", type logical}}),
Deduped = Table.Distinct(Typed, {"eventId"})
in
Deduped
For scheduled refresh in the Power BI Service, install the on-premises data gateway on a machine that can reach the share.

2. Add sessions and measures
Sessions = FILTER ( Events, Events[type] = "LOGOUT" )

Hours = DIVIDE ( SUM ( Sessions[durationSec] ), 3600 )

Session Count = COUNTROWS ( Sessions )

Avg Session (h) = DIVIDE ( [Hours], [Session Count] )

Open Sessions =
VAR Started = CALCULATETABLE ( VALUES ( Events[sessionId] ), Events[type] = "LOGON" )
VAR Ended = CALCULATETABLE ( VALUES ( Events[sessionId] ), Events[type] = "LOGOUT" )
RETURN COUNTROWS ( EXCEPT ( Started, Ended ) )
Add a Date table and relate it to the date of sessionStartUtc in Sessions, so slicers and time charts work by day, week and month.

3. Suggested dashboard pages

Page	Visual	Fields
Usage overview	KPI cards	Hours, Session Count, Avg Session (h), Open Sessions
Chargeback	Stacked bar	Hours by costCenter, split by division
Projects	Matrix	project rows × month columns, Hours values
Sites and testers	Bar or map	Hours by location, drill down to host
Trend	Line	Hours by week, one line per division
Shared accounts	Table	Filter guruUser = Engineer, break down by location and host
Data quality	Table	Rows where mapped = false, and AGENT_START/AGENT_STOP per host

4. Report ideas

• Monthly chargeback: hours per cost center and project for the month, exported to Excel or PDF for finance. Use a paginated report or a subscription in the Power BI Service to email it on the first of each month.
• Tester utilization: hours used ÷ hours available per host. Testers well below the fleet average are candidates to move or share.
• Peak demand: a heatmap of sessions by weekday and hour, to plan maintenance windows and shifts.
• Coverage check: hosts with no AGENT_START in the last 7 days are not reporting; check their Autostart Agent and FTP settings.

5. Tips

• Always remove duplicates on eventId. A retried upload can deliver the same events twice.
• Group users with an upper-cased copy of guruUser, because Guru keeps the case as typed (Engineer and engineer are one account).
• Timestamps are UTC. Convert to local time with DateTimeZone.ToLocal or a per-location offset before reporting by hour.
• A session that runs past midnight counts on its start date when you relate Sessions on sessionStartUtc. Split long sessions by day if daily totals must be exact.
• Count only LOGOUT rows for hours. A session still in progress has no duration yet; show it with Open Sessions instead.
• Fix mapped = false rows by adding Map rows on that tester. Only later events pick up the change, so fix gaps early.