HI @reneboer and @therealdb thanks for your advice ! For the near future I have several options I consider: stay with Vera and new Ezlo controllers in combination with MSR or the openluup option or leave vera/ezlo controllers completely and switch to openluup in combination with zway or home assistant+MSR or even go to Homey.
But for now I stick to my current setup, don’t have the time now to make big changes. I also use The Home Remote app and this integrates my Vera controllers and a Fibaro Homecenter 2 (that I use for some zwave devices that don’t work well with Vera and never will work witth the new Ezlo I fear…) very well.
So what could I do to prevent the luup reloads and controller going online.
What do you think will be the main cause looking at my setup and the logs ?
Try the beta. Luup reloads are usually associated to too many plugins or scenes running together, so there not too many things you could try, beside running them on openluup.
Okay I will try the Beta.
Found the link to 7.32 beta (open Beta 4). Upgrade went well. Curious if the luup reloads will decrease now.
I still have a quite busy Vera plus and it’s doing 4-5 days of uptime, versus 4-5 hours before.
This early morning I again had three luup reloads: 5:14 5:15 and 6:32 This looks like a daily pattern and not changed by this beta
So, what’s your unit doing at that time? Which plugin are running? This is the only way to try to understand what’s going on and try a fix.
Agreed that the current (beta) firmware didn’t stop my Vera Plus restarts; it just spread them out over longer spans. They still happen, about once a week. And that’s on a unit with almost no plugins, no tasks, no Scenes. Everything is managed in MSR.
Luup engine is so full of reloads to prevent hangs, that they are inevitable. I think it’s impossible to completely remove them.
Figuring out what’s trying to go on at that time is the key first step, as @therealdb said. Since deadlocks and crashes may happen a minute or so after the events that cause them, you may need to look backward a good bit from the moment of the crash to see what it’s been doing.
The usual methodical approach here (at least for me) is to identify what’s been happening leading up to the crash, and start eliminating those variables. Start removing those devices from the mix – if Reactor has been running an activity, for example, remove all the actions for one plugin’s devices from the activity (e.g. RFXTRX; leave comments in the place so you know what/where/how to replace them later). Then let it run a couple of days and track changes in behavior. If it still crashes, go on to the next until you get a behavior change, and at that point, you may have identified your problem device/plugin.
Time consuming, but this is the way if you want to get to the bottom of it.
@therealdb @rigpapa thanks for your advice.
I did some checks. The luup reboots at 5:14 and 5:15 have to do with a Reactor scene that I run every morning at 5:00 to clean up memory, followed by a 10 min delaty and then a reboot. Why it reloads twice within a minute I don’t know.
Further I don’t have many timed Reactor scenes at this Vera. Every day at 6:30 i retrieve the weatherforecast for that day (Wunderground data via SiteSensor). So may be the luup reload at6:32 has to do with that ?
Every day when we wake up (usually at 7:15 Iwe get the weatherforecast out of a Sonos Play1 in our bedroom and the bedroom light slowly lights up. But that never leads to a luup reload.
At midnight exactly 0:00 some Reactor Scenes are run to write the daily power, gas and water usage to a Multistring device. It also calculates in % how much the difference is versus 5year average. But also around midnight I don’t see luup reloads.
The smartmeter plugin gets power and gas data every 5 seconds. The actual weatherdata are retrieved every 3 minutes 24/7 The weatherforecast also every 3 minutes but only between 6:30 and 9:30.
I now have changed the time of the weatherforecast scene from 6:30 to 6:50. Curious if the luup reload is now also 20 minutes later.
Question: how can I record the Vera logs in a certain time period longer than 1h ago ? Because the logs go back maximum 15-20 minutes as you check them via http://vera-ipadress/cgi-bin/cmh/log.sh
I don’;t mind some luup reloads, as long as the Vera does come back online every time.
Just my two cents’ here, but I would strongly consider lowering the frequency of those data pulls, based on my experience that Vera reboots seem to correlate less with discrete events than they do with “collisions” of multiple events being executed in rapid succession.
If you plot multiple, regularly-spaced events on a timeline, so long as they involve different intervals, there will invariably come a point in the day where they all “collide”. And Vera, as we know, hates to be caught with her pants down (e.g. still waiting for an IGET or DELAY() to be fulfilled when another request enters the queue).
I agree with you. I would like to lower the frequency of the smartmeter plugin, but this is every 5 seconds and can’t be changed if I remember well. @reneboer ?
OK, this would have been good info to have up front. At least one reload/reboot is explained then. The question then is, how do you know you are getting two reloads? That is, what are you seeing that is telling you it’s reloading a second time after the reboot?
That’s a good move.
By default, logs rotate every 12 hours or when their size hits a threshold (determined by available disk space), so if yours are rotating that fast, you may be logging too much for some reason, or your storage free space is running low. What’s the value of LogLevels in your /etc/cmh/cmh.conf file on the Edge? Default is 1,2,3,4,5,6,7,8,9,50, and I usually remove 6 from that list because it’s just too chatty (reload Luup after making changes). While you’re in there checking, do a df -kh and report those results here; you may be so critically low on disk space that that alone is leading to some reloads.
I have a Reactor scene that sends me a (Vera native) alert every time luup reloads
Screenshot of part of this cmh.config file:
I see rotation interval = 15*60 seconds so only 15 minutes. How can I change that to the 12 hours you mention ?
OK. Look in /etc/cmh-ludl and see if there are filenames for plugins you don’t recognize or no longer use. When you uninstall a plugin, Vera does not remove the plugin files. Also, look to see if you have any files named both with and without the .lzo suffix. This is wasteful duplication; don’t do anything to them yet, let’s just see if you have any.
That rotation interval in /etc/cmh/cmh.conf isn’t used in the rotation script, so I’ve always just ignored it. If you run the rotation script manually Rotate_Logs.sh --force on the Edge, it will tell you its configured interval and disk space allowance. This script is actually run every minute by Luup (without the --force option), and it decides on its own when and what to do, if anything.
I checked this directory and don’t see any filenames for plugins I don’t use except for some that Vera/Ezlo put in as standard (Alexa, ezviz camera).
I also don’t see filles named both with and without the .lzo suffix.
This is the output on my Edge:
And yes… I see a luup reload this morning at 6:51 So this Reactor scene causes a luup reload. It retrieves the 5 day weather forecast from WU. It’s quite some data. Because in SiteSensor you can only define 8 variables I made 2 Sitesensors one for day forecast and one for night forecast. Both call the same URL at the same time. Maybe there it goes wrong. So I put a delay of 30 sec between the “day” and "night"sensor to set them to armed. Will see tomorrow morning