I was thinking of a nice feature to offer.

Once in a while a node gets stuck. It does not update data, so everything seems normal, and it may be ab while before it becomes noticed

A way around this is a heart beat generated by the node (reset a time, so it does not run out on the ISY - it takes several small programs per node)

Recently, I have started to include last update time (uom=151 - prints seconds since 1980 as nicely formatted text string) - this variable can be compared to time-now from the ISY (using a separate isy program) and if difference is larger than expected node update time, a warning can be sent

It would be really cool, if there was an option to automatically restart the node from an ISY program. It cannot be a call to the node, as it is not responding, so it has to be an option to restart the node using the ISY

Restart the Plugin?

Plugin that restarts other Plugins

Or a function within eisy-ui

14 hours ago, Javi said:

Restart the Plugin?

Plugin that restarts other Plugins

But what restarts the plugin that restart the plugins

47 minutes ago, apostolakisl said:

But what restarts the plugin that restart the plugins

Another instance ,

Joking aside, for node servers which are command only, like iTach, I have not seen them disconnect. Unless there is an issue requiring reboot or reinstall which a restart would not correct anyway. These sort of things are the reason we are moving to "plugins" from "pg3", now being native this method would likely be the best route although would require a permission. I'll inquire in our next meeting.

I have seen nodes hang - some times errors caused by the FW can make some nodes hang - clearly they can loose connection to the data source as well but that is also covered if we use last update as a parameter to drive a restart

Our team reviewed the request and a supervision-plugin would be the most "eisy" way of allowing control from within programs. Initially this would be the ability to restart from a program, but could be expanded to included plugin status/start/stop/etc. ID: /udi-tasks/issues/16 created however we are currently may weeks out on feature requests.

With that in motion, there is also an obscure property id "ERR" named "Responding", (it is inverted by name as implemented long ago), it used in all native families. This property is used to show issues on individual nodes with an error badge in the admin/eisy-ui node list. In native families this is not set by the node itself, so there may be a way to implement a "heart-beat" without the need for users to track elapsed time, this could be triggered by the plugin software. This is still an early investigation, but I hope to review in more detail when time permits.

I think this is a good start - however, it needs to be from an external node - if the node hangs, it cannot be expected to do anything like testing for a status

My 2 cents

If it fails we have a perfect test case, all internal apis.