This is with a Vera2, running 1.1.1362. But I saw the same behavior with 1.1.1338–upgraded in the (silly) hope that this might have been one of things that had been addressed.
I am pretty sure this has to do with my darned HRSD1s. I have four of them (and would love to replace them with something better). One is a direct hop from Vera, and that one seems pretty stable. I am not seeing anything like the battery life I used to see from these things with UI2, and suspect it may be due to some drastic change in the polling strategy MCV is using for them in later firmware versions.
The other three are out in my garage, and require a hop. When I first moved to the V2, I added these and they were fine for a day, and then Vera exiled them on the first “automatic nightly heal”. I opened a trouble ticket, turned on tech support, things went for a month or so, no real change, occasional mail from MCV. I left on vacation for three weeks, and when I came back, decided I would attack them myself. Removed them from the network, re-added, re-configured, set up manual routes for them, and set them not to be configured by Vera. Everything worked fine for about a week. I disabled tech support, as I had heard nothing from MCV in a month, and figured my stuff had gotten shoved to the back burner and had fallen off the stove.
While on vacation, I had become annoyed with my existing Square Connect setup–my device names were a bit too long and too descriptive, and were spilling out of the SQC widgets. So the other thing I did when I got home was a fairly substantial amount of device name shortening. After that was done, I found that Vera seemed confused about communication with some of the renamed devices, and started showing stale state for them. This may be unrelated to what I am seeing now, but it was the first sign of trouble. It could also have been that my HRSD1s were beginning to thrash the system. Dunno.
Two days ago, two of the garage sensors went from showing reasonable status to showing stale battery readings. I have replaced the batteries in one (I am using Eneloops, but if the darned things are only going to last 4-5 days, am not sure what I’ll do), and will replace the other’s as soon as I have a set of three batteries get reconditioned and charged.
Last night, the dusk scenes failed to execute. I ran them by hand, and they started everything up fine, but the state of the devices in UI4 didn’t show that consistently. We went to bed, and the “lights out” sequences failed to execute. I had noticed the frequent “server busy” popups, and last night had figured one of them had the system tied up when the scenes should have started.
Closer examination this morning shows that I am really getting 15-20 seconds of “server busy” at the top of each minute. I am guessing that this pretty much torpedos any scene fired off a timer, since I currently have those starting on even minutes. It also seems to be keeping Vera from ascertaining correct device status. I can turn a device on or off that is showing stale status, and the UI will reflect the change (and the change happens), but then the “server busy” even happens, and it looks like because polling got blocked or something, the UI reverts to showing the old, stale status it was showing before.
Highly annoying, and makes the system nearly useless. I have added a 16G USB stick to the system, enabled logging to that and enabled verbose logging. Sure as the world, I am seeing a hailstorm of chatter that starts right at the even minute and runs 15-20 seconds, and then things go pretty quiet. I am seeing something in each pile complaining that node 36 is “bad”…it’s a Leviton fan controller that I renamed but had to delete, re-add, and reconfigure before it started showing accurate status. Vera seems to have concocted some pretty fanciful “auto-routes” for it–the device is perhaps 8 feet away from Vera. But that device is now showing a “1” for CommFailure, so perhaps it’s contributing to the trouble.
Don’t know. Has anyone seen anything like this? I have seen the odd post from Florin, but doubt that MCV is going to get back to me before next week, even if I open a trouble ticket.
–Richard