This evening about 10pm my ZWave Vera restarted and as it doors about once a week LuAUPnP DID NOT START. How can one trust this thing if it keeps doing this at random.We travel quite a bit and I need to know that things continue to work while I’m away.
It answers the Ping and I can access via SSH. So a ping controls automatic power switch won’t work. And when I SSH I can see what’s happening but it doesn’t respond to a REBOOT command. Very frustrating and Tech Support never seems to have an answer to this situation.
I know all too well your frustrations. Vera, even with the latest update is TOTALLY unreliable. My unit adds devices all by itself, fails to run timed scenes, and when using the web interface, EVERY browser I have tried gives me “not responding” messages for minutes at a time. Vera should have never been placed on the market… It is like something someone would make in their garage, not a retail device at all.
Mine is nowhere near that. Mine works properly about 99% of the time. It’s slow sometimes but generally good enough.
But what makes me crazy is that 1% of the time when the damned LuaUPnP fails to start and no one at Vera can tell me a) why this happens, b) how to recover from it short of a manual power cycle and c) what can be done to stop if happening in the future.
How hard could it be for the Vera to detect it and force a restart if it falls to start properly? Or better yet code it do it doesn’t fall to start.
I wonder if the NetStatusID flips to something other than 1 on the Zwave device when this occurs? I created a virtual panel to see the current status of the Luup engine.
I don’t know where to go from here. Vera is just frustrating.
For the last several days I find PLEG schedules not executing. I have a schedule to the family room that prevents the lights from going off without motion after 6PM and before 4AM. for three days now it hasn’t executed. The living room schedule turns on the lights at sunset+1 hour and off at 01:00. Yesterday it didn’t execute and today I found it still false. The master bedroom schedule prevents light from responding to motion after 10:30 and before 08:00. Two days now it has executed at 10:30 and not shut off as scheduled at 08:00.
I suspect it has to do with the ridiculously high CPU utilization, never less than 25% and most of the time over 80-% with frequent spikes to 95% schedules don’t get serviced. Lights take 20-45 seconds to come on when triggered by motion sensors or door sensors. I’ve been through the system with a fine tooth comb trying to eliminate any obvious problems but to no avail. I do have lots of timers running in PLEG to replace CountDownTimers to turn lights off after periods of time and I’m beginning to think this made it worse not better.
I’ve tried removing AutoVera, ImperiHome, EventWatcher et al and it goes down a little but I’ve lost lots of functionality in the process.
My UI5 Vera has about 45 devices running from zwave switches and dimmers to camera’s and about 10 plugins running. Ranging from the Ergy one to the Yamaha and YamahaHTTP ones. About 30 or so scenes as well.
Memory availability is between 95 kb and 102kb. CPU load between 0.19 and 0.31. She last did a CMH et all restart the 19th of this month. Before the the last one was early January.
She’s pretty much rock steady.
My UI7 Vera has about 27 devices running. Mostly battery operated zwave sensors, camera’s and the MyQ garagedoor.
She’s running about 8 plugins ranging from the Blue Iris one to the Virtual panel ones. Only about 10 scenes so far.
The reason she has all sensors is that those were purchased after I upgraded that Vera to UI7 and I was running it as a QA box. It turned out to be stable enough for me to pull it in productions and start added all new Zwave devices to it rather than the Ui5 one. I’ll soon be starting to move some dimmers/switches over to see hwo that works.
The last CMH et al restart was this morning. Before that a few days ago. But then again, I use her for tinkering with the rstrouse’s Virtual Panels. CPU usage is a bit higher between 0.21 and 0.38 but one virtual panel is shows a lot of Vera status variables and that seems to take some cycles. Memory usage is also higher as I only have between 73kb 91kb available.
Not as rock steady as my Ui5 unit but not bad either.
The only difference I can see is that I do not run PLEG. I had it installed a while back and never made much use of it so I uninstalled the plugin.
I have about the same device load (UI5) but the only time I’ve seen single or Lorre double digit CPU loss is in the first minute or so of a cold start … Within minutes it’s above 80pct and rarely drops below 75pct. I’m just completely frustrated
Do you have any serial devices? I have found that I can cause Luup to fail if the device is communicating when it is starting. My alarm panel does this because the plugin enumerates its children on startup while it initialized them. One of these days I am going to move the child init code out of the chdev loop that creates the child devices.
The only other thing that has sent my Vera into unpredictability was a Linear switch with an association that just kept sending flood commands over and over. The alarm panel thing could be cleared by initiating another restart but Vera would be unresponsive for a few minutes waiting for the cancel serial cancel. The association problem required a full on restart and it would fail seemingly without any reason. It drove me insane for a while and I happened on it by accident while trying to associate another switch.
I don’t know if any of this helps but maybe you have a similar scenario.
Running pretty solid. Does not reboot. Loop restart if there are problems with internet access (had some issues with this lately but installed a new modem yesterday that should resolve this).
29 z-wave devices
9 Arduino node devices (containing a total of 19 sensors or actuators)
17 RFX devices (containing a total of 27 sensors or actuators)
[quote=“BOFH, post:6, topic:186459”]My UI5 Vera has about 45 devices running from zwave switches and dimmers to camera’s and about 10 plugins running. Ranging from the Ergy one to the Yamaha and YamahaHTTP ones. About 30 or so scenes as well.
Memory availability is between 95 kb and 102kb. CPU load between 0.19 and 0.31. She last did a CMH et all restart the 19th of this month. Before the the last one was early January.
She’s pretty much rock steady.
My UI7 Vera has about 27 devices running. Mostly battery operated zwave sensors, camera’s and the MyQ garagedoor.
She’s running about 8 plugins ranging from the Blue Iris one to the Virtual panel ones. Only about 10 scenes so far.
The reason she has all sensors is that those were purchased after I upgraded that Vera to UI7 and I was running it as a QA box. It turned out to be stable enough for me to pull it in productions and start added all new Zwave devices to it rather than the Ui5 one. I’ll soon be starting to move some dimmers/switches over to see hwo that works.
The last CMH et al restart was this morning. Before that a few days ago. But then again, I use her for tinkering with the rstrouse’s Virtual Panels. CPU usage is a bit higher between 0.21 and 0.38 but one virtual panel is shows a lot of Vera status variables and that seems to take some cycles. Memory usage is also higher as I only have between 73kb 91kb available.
Not as rock steady as my Ui5 unit but not bad either.
The only difference I can see is that I do not run PLEG. I had it installed a while back and never made much use of it so I uninstalled the plugin.[/quote]
All–
I have 3 different veras 3 running UI 1.5.622. No problems with UI5. My experience is like BOFH’s -“rock steady”. I do run PLEG and can not identify any problems.
I am going to assume you already restored factory default, re-upgraded to the latest UI5 image, and rebuild your zwave network.
Before I start pointing fingers at the actual Vera hardware, I would try replacing the power adapter. Any MIPS SoC board can be flaky when not provided stable power. If that doesn’t solve it, sorry, you’re probably looking at a hardware replacement.
Other that that, I can say my UI5 Vera was rock solid before I upgraded to UI7. It took all of 48 hours of constant zwave network loss and device pairing issues before I went back to UI5 and all my problems disappeared.
This morning I awoke to find that none of the overnight scheduled had executed, that is lights that turn off in the early hours of the morning were still on. I logged into the Vera and found the CPU at a mere 15pct … But memory at 235 pct, yes 235! That’s the first time I’ve seen that! I’m accustomed to memory usage less than 80pct and CPU consistently above 90 pct so that was new and probably the root cause why nothing was working as scheduled. EVERY DAY SOMETHING NEW WITH VERA.
This is UI5, 1.5.672 to be exact for the Fibaro motion sensor
I have about 45 ZWave devices attached and the following Plug Ins
PLEG 5 instances
Event Watcher logs to a SysLog on the LAN
AutoVera
ImperiHome 3 instances
Aeon Home Energy Monitor
GCal3 calendar sensor
Day/Night
Vera Alerts
MultiSwitch 5 instances
Yesterday I took a backup and then started removing Plug Ins while monitoring over SSH. I thought I had hit pay dirt when I removed AutoVera and ImperiHome and my CPU was in the single digits … For about 5 minutes and then jumped back up to 80pct and stayed between there and 95pct just like always. I eventually had most of the Plug Ins removed and it never got better for more than the first few minutes after restart. Currently it takes anywhere between 15 and 45 seconds for a light to go on in response to a motion or door sensor.
The fact that it takes several minutes before it tanks makes me think it has something to do with the timers and schedules in PLEG. I have a lot of timers in PLEG but of course most of them are self-retriggering and used for lights on/off, etc. so they are idle until a light goes on, door opens, etc. There are a few absolute interval timers that of course start when PLEG starts and some schedules, most of which are active in the evening/night.
[Update]
I just took another backup and removed ALL the PLEGs. After I removed the final one Vera restarted and via SSH/TOP the utilization showed in the single digits — for about 1 minutes and then BANG, back to 80+pct. So its not the PLEGS that are causing it.
Vera writes out it’s state every 6 minutes.
Sounds like and could have failed and reverted back to a previous use state.
Do you know if Vera restarted during that time frame ?
Having a timer in PLEG does not have any runtime cost (a very little bit of memory).
What uses CPU is when PLEG has to evaluate conditions:
a) An Input Triggers
b) An Input Device Property Changes
c) An Input Schedule is started or stopped.
When you uninstalled the ImperialHome plugin … did you also stop the tablet apps ?
How many REMOTE apps do you have running at one time ?
Each REMOTE apps has to setup a polling loop to Vera … And to support this Vera has to keep track of what it sent to each remote app.
This will use memory AND cpu and will grow as the # of remote Apps continue to increase. A BROWSER with the Vera Desktop displayed is ONE remote. ImperialHome, HomeWave, Vera’s Mobile App, OpenHAB, … are all REMOTES.
You might try limiting the remotes to one at a time …
I forgot, my UI5 also has 5 Android tablets and phones connected running AuthomationHD which I use as control panels in the various rooms.
Since Clippermiami has been removing plugins and the issue does not go away, I’m starting to also wonder about the hardware and power. Both my Vera’s are on the same power conditioning heavy duty APC UPS that also services my cable modem and main server.
We had a wierd one at work a while back with a Fiery controller attached to a copier. It would just freeze. IT worked with the copier techs and no luck. Until the low endUPS it was on broke and I put a heavier one in there that had a voltage display. Sure enough, one day when checking the Fiery, I noticed it reading 129v and watched it for a bit. It starting going up and down between 129 and 118 for about 10 minutes and then went normal again. We had property management check the circuit and sure enough there were issues. They ran an new mains run from a different circuit and the Fiery has been working great ever since. The copier itself is still on the old circuit but it seems to be able to deal with it just fine.
Perhaps run a multimeter on the Vera PSU output and see if the voltage is within spec under loaded conditions? at this point, eliminating what is not causing the issues should leave clippermiami with what it is…
Eliminate all of the following:
Cameras
AutoVera
ImperiHome 3 instances
Aeon Home Energy Monitor
Cut down REMOTE APPS to 1 instance (I suggest it be a browser during this test)!
REBOOT (or Power Cycle)
REFRESH Browser
Get a baseline … start adding things back.
[quote=“RichardTSchaefer, post:15, topic:186459”]My recommendations would be to:
Eliminate all of the following:
Cameras
AutoVera
ImperiHome 3 instances
Aeon Home Energy Monitor
Cut down REMOTE APPS to 1 instance (I suggest it be a browser during this test)!
REBOOT (or Power Cycle)
REFRESH Browser
Get a baseline … start adding things back.[/quote]
Here’s the latest.
There were never any cameras on this Vera.
I removed
— AutoVera
— Energy Monitor
— EventWatcher
Still attached are three Imperihome tablets, the browser and “Juice” SSH Android utility. All the other devices are still attached along with all the PLEGs, MultiSwitches and Vera Alerts.
In the past 2 hours the nominal CPU usage is 30% and I’ve been watching it for the past hour (BORING ) has only been above 50% once and that for a single SSH/TOP cycle. So it appears that whatever was driving it to 90% consistently is out of the system. The response to motion/door sensors is back to excellent. So that appears to be the benchmark.
Tomorrow I will try reinstalling AutoVera and the Aeon Energy Monitor and see what happens.
Thanks to everyone who has piped in. Again my apologies for venting but as some of you have experienced the frustration level with Vera can get very high very quickly.
My UI5 Vera 3 experience fI had big unstability issues when I used AutoVera. My Vera would restart several times a day and devices were very slow. I really needed Autovera, but once I uninstaqlled it, my Vera is stable with no memory issues. It restarts only during the nightly heal.
I waited a few months and installed it again. Within hours it went back to unstable operation.
my frustrations with vera are documented in my posts on this forum. I am using vera at a vacation home, while I am sticking to my 15+ year old X10 installation, 2 Nest thermostats and an ATT DigitalLife security system for my main home.
so I have been looking at alternatives to vera for a while, but I have not found a z-wave system that - according to reviews and their respective user forums - does not have issues. I guess at this point one can only wait for Apple, Amazon or Google to bring to market a reliable home control system.
which is a shame, because Vera was initially in a strong position to lead in this market. but UI7 buried all my hopes. and I’m not ready to go thru the hassles of downgrading to UI5.
) has only been above 50% once and that for a single SSH/TOP cycle. So it appears that whatever was driving it to 90% consistently is out of the system. The response to motion/door sensors is back to excellent. So that appears to be the benchmark.