Theory On Why So Many Unexplained Vera Restarts

Plugins & Plugin Development General Plugin Discussion
1

If the Tunnel to MCV is established … All events are sent to the MCV Servers.
What if the MCV server is slow ?
Does this cause backups and overrun restarts on Vera ?

If you generate a lot of events … but have plenty of memory … this may be the problem.

2

I looked at this earlier in the yr, as MCV had left my machine unstable for 4 months after the uptake of a private Beta release (2-10 restarts/day consistently). This release was eventually pushed out as the 621 release.

At that time, the biggest cause of my restarts was a faulty [tt]NetworkMonitor[/tt] process. I had published data/analysis showing that it was artificially killing off the LuaUPnP Process, after it considered it to be non-responsive. At that time I added separate [non-Vera] monitoring of the LuaUPnP to show that it was working perfectly, when the [tt]NetworkMonitor[/tt] process was seeing it as non-responsive (to HTTP ping calls)

An interesting side-note was that if it didn’t respond to one-ping, then subsequent pings would also fail (even though they passed when executed by my Mac)… something that might happen if [tt]NetworkMonitor[/tt] didn’t handle bad connections correctly, for example.

In my case, it was this behavior that consistently made my Vera unstable, often being restarted during the more critical of my scenes (again, responding correctly to my Mac on HTTP Pings, but being restarted by NetworkMonitor for one “miss” in their repeated calls)

Since then, I’ve made many changes to Vera [3], including removing the MCV Servers (completely), substantially decreasing the log levels, removing excess plugins, and removing/repairing all Z-Wave nodes that indicated problems of any sort (mostly, the battery stuff, but a few odd-ball devices also)

After this combination of changes, it’s been relatively stable.