zieren
commented
2 years ago The background is a user report about the client "not working".
Essentially the client is already logging activity to the server all the time. As long as that is happening everything is OK, so we should focus on detecting if the client is ever NOT running.
One idea is to do this on script startup:
- Compare the last shutdown time (available in the registry) with the most recent activity log, which shouldn't be more than a few minutes older.
- Compare the startup time with the current time to ensure it is no more than a few minutes old.
However, users may start the machine long before logging in, and vice versa when logging out and shutting down. So what we really want is user login/logout time, and that doesn't seem to be available in standalone machines.
Maybe it is actually easier to run a second script to monitor that the actual script is running? (example)
Also, this doesn't address the question of identifying the reason for the failure. We should probably enclose the main function in a try-catch block and also monitor time since the last successful request. Then log to a file (or to the server?) to capture any issue that either throws in the main loop or leaves the last successful request timestamp unupdated. Maybe use OnError()?
The client should log recent operations. (As long as the server is reachable it might even log remotely.)