33.1 Acinta Server¶
Acinta Server is a Windows service that works when nobody has Datashop open: it sends distributions on schedule, refreshes materialized dynamic views, and produces the pages for Datashop Web. Older texts call it Saturn Server.
The Service¶
| Name in Windows Services | Acinta Server (service name SvcSaturnServer) |
| Installed by | The server installer – see 30.1 Installing Acinta |
| Start and stop | Windows Services, Enterprise Manager's Server page, or net start / net stop SvcSaturnServer |
| Settings | AcintaServer.msx in the server's program folder |
| Projects | projects.dspx in the server's program folder |
| Log | C:\ProgramData\Acinta\AcintaServer Log.html |
The Service Account¶
The service runs as LocalSystem unless you choose an account. When the data warehouse uses Windows authentication, run the service as a domain account with read access to the data sources and read/write access to the project repository: in Windows Services → Acinta Server → Log On.
Which Projects It Serves¶
The server loads the projects marked as enabled in its settings, at start. If an enabled project can't be loaded – e.g. its database is unreachable or its version doesn't match – the server stops. Check the log.
[!WARNING] Saving Project and mail settings keeps only the project chosen under Active project enabled; all other projects are disabled on the server. With several projects, check the settings file afterwards.
The project must be upgraded to the server's version – see 30.3 Updating Acinta.
When to Restart¶
The server reads its settings only when it starts. Restart it after:
- changing mail or server settings;
- updating Acinta or upgrading the project;
- changing distributions, if they don't run as expected – Datashop reminds you when you save a distribution.
What It Does¶
| Task | See |
|---|---|
| Sends distributions on their schedules, with status reports | Chapter 18 |
| Refreshes materialized dynamic views (when Run materialization schedules on Saturn Server is on) | 27.2 |
| Takes the daily system health snapshot (build 20238 and later; called database health snapshot before build 20348) and cleans up the audit log | 32.3 |
| Records each distribution run in the audit log, and keeps a session open there so Datashop and Enterprise Manager can show whether it's running (build 20230 and later) | 32.3 |
| Produces Datashop Web pages and sessions | 36.1 |
Synchronization¶
Build 20265 and later.
Acinta Server saves at the project's Automatic synchronization/save interval (30 seconds by default), and each time loads the changes users have saved in Datashop, Modeller and Enterprise Manager – see 20.4 Saving and Working Together. It's on by default; <SynchronizeMetadata>False</SynchronizeMetadata> in AcintaServer.msx switches it off.
While it loads changes, new report requests and scheduled actions wait until loading is done – usually only a moment. If reports or actions are already running, the server waits up to 60 seconds for them. If they are still running then, it leaves the changes for later and tries again after 5 minutes. Nothing is lost.
What users notice: a report may start up to about a minute late just after someone saved changes, and on a very busy server changes can take a few minutes to appear.
Safety Stops¶
To protect the server, the service stops itself if its log grows by more than 5 MB in 10 minutes, or the disk has less than 100 MB free. Check the log when the service has stopped unexpectedly.
Build 20334 and later: Acinta Server doesn't wait forever on a database: when it reuses a database connection, it checks that the connection still works and opens a new one if the check gets no answer within 30 seconds. A scheduled run waits at most 2 minutes to write its row in the statistics database; then it continues without it and writes a warning in the log. Loading a report's data is limited by the project option Time limit for loading report data (minutes) – see 26.2 Project Options.
The Status File¶
New in build 20349.
Every 30 seconds Acinta Server writes a status file: C:\ProgramData\Acinta\Server\<computer name>.status.json. The log names the file when the service starts. It's meant for monitoring tools, but you can open it in any text editor. It shows:
- when it was written – if the time is old, the server has stopped or hangs;
- the server's state, version, start time and loaded projects;
- the process's memory, threads, running jobs and the number of errors logged;
- the number of schedules and the next run;
- running and recent distribution runs, with e-mails sent and failed, and how many scheduled runs were skipped since the server started because the previous run was still running – see 18.8 Status Reports and Troubleshooting;
- the computer's memory, CPU and free disk space.
The health section sums it up as ok, degraded or stuck, with the reasons in plain text:
| Health | When |
|---|---|
stuck |
The scheduler has been silent for more than 15 minutes, or a report has been loading data more than 5 minutes past the project's Time limit for loading report data (minutes). |
degraded |
A distribution has been running for 12 hours or more, a report couldn't get a database connection in the last 15 minutes, free disk space is below 2 GB or 5 %, or free memory is below 10 %. |
With stuck, the watchdog restarts the service (build 20356 and later) – see The Watchdog below; otherwise restart it yourself and read the log.
The Watchdog¶
New in build 20356.
A second Windows service, Acinta Server Watchdog (service name AcintaWatchdog, AcintaWatchdog.exe in the program folder), keeps an eye on Acinta Server. The server installer installs and starts it together with Acinta Server – also when you upgrade, including from a version without a watchdog – and removes it when you uninstall.
Every minute it checks the Acinta Server service and its status file (see The Status File above). It restarts Acinta Server when the service is running and:
- the status file hasn't been updated for 5 minutes – the server hangs completely; or
- the status file says
stuck.
A restart is a normal stop through Windows. If the server hasn't stopped after 2 minutes, its process is ended. Then the service is started again.
The watchdog does nothing:
- while the Acinta Server service is stopped, paused, starting or stopping – a stopped server was stopped on purpose, and the watchdog never starts it;
- in the first 5 minutes after the service has started;
- when there's no status file (Acinta Server older than build 20349);
- for
degraded– that never causes a restart; - after 3 restarts within an hour – it logs an error and waits until the hour has passed.
Each restart is logged as an error, with the reason, in C:\ProgramData\Acinta\AcintaWatchdog Log.html. The watchdog writes nothing to the Windows event log and sends nothing outside the computer.
To turn it off, stop the Acinta Server Watchdog service in Windows Services and set its startup type to Manual or Disabled.
The installer also turns on Windows service recovery for both services: if one crashes, Windows starts it again after 1 minute. The failure count is reset after a day.
Troubleshooting¶
Problem: Distributions aren't sent. - Solution: Check that the service is running, the project is enabled in the server settings, and the log for errors. Restart the service after changes. With the audit log on, the Last run column in Datashop and the distribution tree in Enterprise Manager show whether Acinta Server is running and how each distribution's last run went – see 18.1 What Distributions Are.
Problem: The service stops right after starting. - Solution: Read the log – often a project can't be loaded (database unreachable, wrong version) or the service account lacks database access.
Problem: The service is running, but nothing happens.
- Solution: Open the status file (see The Status File above). If it hasn't been written for more than a minute, or health says stuck, restart the service and check the log (build 20349 and later). From build 20356 the watchdog restarts a hanging server after about 5 minutes.
Problem: Acinta Server restarted by itself.
- Solution: Look in the watchdog log, C:\ProgramData\Acinta\AcintaWatchdog Log.html – each restart is listed with its reason, e.g. the status file wasn't updated, or the server was stuck (build 20356 and later). Then read the Acinta Server log from just before the restart to find what hung. If the watchdog log has no entry, the server probably crashed and Windows service recovery restarted it. The Windows event log only shows that the service stopped and started again; the Acinta Server log may show what it was doing.