joshsoftware / minion-frontend Goto Github PK

Right now the forgot-password functional are unimplemented. There should be a viable implementation built.

Setup a basic kanban board to visualize work and issues in this repo.

There are two scenarios that need to be, in general, handled gracefully, and that are not right now.

One is if the API server doesn't handle the request at all. This generally passes without notice, except for error messages in the console, but this is almost by accident instead of design. Proper error handling needs to be added.

Second, if the API server has an internal error, it will return an appropriately sized nil set response. The front end needs to detect when it gets nils instead of data that it was expecting, and cope in a reasonable way.

Particularly for longer interval cycles, it might be nice to have some sort of subtle indicator showing when it will poll again.

This page needs at least a skeleton so that it works, and real content can be added.

Right now the count totals are global record counts, instead of counts pertinent to that specific user.

If the log selection criteria changes, a full new set of logs is pulled. When that happens, the existing timer should be cleared, and a new timer started, since the adaptive polling timer will need to look at the new log set to determine polling frequency.

This page needs a skeleton so that it works and real content can be added later.

With all of these fancy tools to search and segment logs, it should be possible to download a copy of what was found.

The service types are populated by an API call when the card is opened. Network latency means that sometimes a person has already clicked on the dropdown to select a service type before the data has been returned.

The simple thing here is to just show a loading indicator if there are no service types. There should never be a scenario except for one in which an organization has been setup, and there is a record for a server, but no logs have been collected, where there are no service types, and that should be very rare, so can safely be ignored at this time.

It is probably sufficient to use the Skeleton loading indicator that is being used elsewhere. This should be simple.

If a query results in no logs being found, the log reload button is not visible. It should be visible because it may be that there are no logs yet, but there will be soon, and the user would like to just poke reload a few times until they start to appear.

We need a login page for regular users of the system.

The User Admin portion of the admin panel is unfinished. Finish it so that users can be properly administered.

There is debug info being output into the loading banner. That should be cleaned up. It's kind of ugly.

Right now it's just user/password. We should support 2fa, and make the use of it a toggleable feature on the account.

The default right now is 500 lines, which seems like a good number for telemetry, but it's too much for logs. Most of that data will be transferred, but never looked at.

It should be something smaller, like 100 lines.

There is still a LOT of carried forward cruft and junk from the template that I used to provide the base widgets and building blocks. This needs to be cleaned up.

Start out with the simplest thing -- an empty user dashboard.

Tooltips should be added to the log polling control buttons so that it is clear what they do if one hovers on them.

This page needs a skeleton so that it works and real content can be added later.

Start out with an empty administrative dashboard.

If one is searching for a specific log entry, one may want to pause polling when one finds it, so that new logs don't cause what was found to scroll. We can add a simple iconbutton next to the refresh button to pause polling. Along with this should be a "play" button to restart polling. Play should be disabled while polling is active. When polling is paused, the pause button switches state to inactive, and the play button to active, making clear their purpose.

When new log lines are added to the table, the dashboard should apply a very brief, subtle flash effect to those table rows to give the eyes something to latch onto to indicate that new data has appeared.

We need a login page for admins of the system

The last major functional piece to need MVP is command invocation and the viewing of the results of previous command execution.

The UI was written with the assumption that we would support Google/Github/etc. OAuth integration. None of this is written, though; it's just icons and pictures.

Either support it, or remove the icons until we decide to support it.

Logs should poll and update automatically, but if there is a long wait between polling cycles, as determined by the interval algorithm, and a person thinks that there should be new logs, a person might want to be able to poke a button in order to see that.

Add a load type button to manually trigger a polling cycle.

This page needs a skeleton so that it works and real content can be added later.

Telemetry does not auto update, because that's just too complicated to deal with right now, and that much polling could get burdensome. So, we should provide a simple reload button so that one can poke it to see updated graphs.

There are probably too many logging statements still floating around.

Waiting 60000 seconds to poll for new data.

Yeah.....that's too long. It was supposed to be 600 seconds -- 10 minutes. That seems reasonable for a log data that just has not had very frequent changes.