Long-term stability of openhab

I would like to know how stable openHAB runs by other users? Do you need to restart openHAB from time to time? Days, weeks, months?

My installtion is not so stable at the moment. After a few days I can see from my charts that some values are updated only sporadically. Or not at all. In this case I have to restart openhab and everything runs again.

I already patched “ComfoAir” binding in the past to prevent the binding from hang on ther serial out flush(). That has worked very well. But now I’m back to the original 1.8 version and the problem reappears.

Here my current setup:

  • Odroid U3, 8GB eMMC on AchLinux
  • 4x Serial2USB converts connected via powered USB hub
  • FTDI, CH340, PL2303 and FTDI fake
  • Weather Station via USB HID
  • Sun Java 1.8.x
  • openHAB 1.8

Used Bindings:

  • eBUS
  • Serial
  • ComfoAir
  • Homematic with CCU2
  • Astro
  • Exec
  • WMR Weather station

After I removed the SONOS binding, the binding became stable.
I am running windows, so due to updates of windows, I have to reboot about every 2-3 weeks, therefore I have no experience for longer periodes.

Running openHAB 1.8 and had been up since the upgrade until I restarted to change Weather from Yahoo to Underground yesterday.

HTTP
InsteonPLM
Mios
ModBus
Mojio
MQTT
Weather

I’m running windows 7, it’s running as a VM on vsphere. 2 cores, 2gb of ram, 50gb HD on ssd storage… nothing special. Hyper visor is a single socket 8 core small business dell poweredge. Things were touch and go on windows server R2 and windows 8.1 even with 4gb of ram but just fine on windows 7. Java and windows updated last week and completely screwed the system. I had to do a system state restore from backups to get it back up and running. For me the OS is the biggest variable and cause for concern.

Bindings used:

  • Astro
  • mqtt (still trying to figure this one out)
  • Squeezebox
  • http
  • networkhealth
  • ntp
  • Onkyo
  • Serial
  • TCP
  • Weather
  • xbmc
  • zwave

I’m no smart home\openhab expert but at the end of the day openhab is a computer system so as a systems engineer let me tell you what I would do. For a low impact system like this with layers you need to peel back to the basics. With no knowledge of the origin of your hardware or your technical abilities i’d start with the following:

  • Remove all configurations (bindings, items, rules, everything)
  • stop openhab and leave it off with the base OS running for a few days
  • Start openhab with no configurations and leave it running for a few days (at the most create a single rule that sets a single variable AND write to the logs one per hour)
  • Add back your binding one at a time and let it run for a few days (I would start with the ones that interact with physical hardware like those usb to serial converters(Wouldn’t shock me if your problem reappears right there))
  • Add back the items file for the previously enabled binding and wait a few days
  • Add back rules for the previously enabled items and binding and let it run for a few days
    You may be tempted to fix any problems that come back as you add something back, don’t do that. remove the failed item, rule, etc and move on the to next one. You will obviously want to make sure that removing the previously added component restored stability before moving on the the next one.
  • Rinse and repeat until everything that didn’t fail is loaded and let the system run for a few days.

At the end if your problem shows itself for only one component then it’s an easy one. If multiple components causes instability in your system, it’s a little harder to preemptively troubleshoot but at least you can come back here with a more specific question. You may notice that every item that failed the test connects to a usb converter or some other common denominator. What ever you do, don’t make changes to rules or configs while you’re waiting a few days to test stability(I’ve had seasoned engineers under me who couldn’t figure this one out!..:disappointed:). There are obvious ways to shorten those trouble shooting steps, If it’s not obvious to you then stick to the list (no offense). Also,this all assumes the logs were not helpful…

Stability of my openhab is affected mainly with stability of the z-wave network. Having problems (not often, once every 2-3 weeks), I implemented two watchdogs: one restarts OpenHab on z-wave stall, the other resets the host when log stops to grow, i.e. at USB sticks disconnects (this happened twice in 6 months).

I’ve never had to restart the openHAB server because of instability. I’ve only ever restarted it because I was upgrading, experimenting, or editing openhab.cfg or logback.xml. I use MQTT, Weather, RRD4J, Zwave, HTTP, Exec, Pushbullet, Network Health, NTP, and a few other bindings which will probably be removed shortly.

1 Like

Thank you all for your feedback. I reduced the number of usb serial converters and wait for any issues. I also modified the binding ComfoAir to reduce the cpu consumption and removed flush() that causes hanging from time to time.

So now my cpu usage is around 5% on an small ARM computer :slightly_smiling:

Running on a RPI2 8gb sd
Attached hardware:
5TB HDD
4TB HDD
FTDI DMX

Software
Oracle java 1.8.x
OpenHab 1.8
Samba
MySQL
Mosquitto
OLA
MiniDLNA

Bindings:
Insteon PLM
XBMC
DMX
MQTT
Network Health

And it hasn’t been stable, although I think I have fixed it now. I was having to reboot it multiple times a day, maybe it would run for a few days if I was lucky. but I modified my tmpfs file and it seems stable, at least for the few hours it’s been running. here is the link to that conversation