today, at ~11:20 AM all persistence services suddenly stopped working and I see absolutely no reason why…
They don’t really stopped working, I can read the data, but no data is stored anymore since ~11:20 AM.
What is strange, I have InfluxDB for some Grafana stuff und also RRD4j for the usual Item graphs inside OpenHab. And both of them stopped saving at nearly the same time. In Log Viewer nothing is visible, no red Lines…
I did some days ago an update in openhabian-config, everything should be at latest stable version:
openHAB 3.4.2 Release Build
Influxdb-Version: 1.8.10
I’ve no idea what it could be but have you tried restarting OH to see if that kicks it off? If not, maybe something is logged in the debug or trace level logs. At least you can see if OH is even trying to save the data and something is going wrong or if the data isn’t even being saved.
Some other things to check include:
are these Items in fact updating?
the add-ons are installed?
check to see if you have .persist files, if so check them for correctness
Can you check if items in /var/lib/openhab/persistence/rrd4j are updated every minute?
You have enabled zram, right? Maybe an issue with zram? Not sure…
I think something is wrong with the latest version of openhab
Restart of openhab service helped again, but if that happens every day, that is not acceptable…
So, I’m pretty sure this is an new Bug in OpenHab 3.4.2
Before that update, at v3.4.1 I didn’t hat that problems, that began after that update, some days later.
After restart of openhab service it work again for some time. In LogViewer nothing is shown.
Items are updated all the time, this is also shown in LogViewer.
Depending on how OH is installed the logs are either stored in /var/log/openhab or $OH_HOME/userdata/logs. When ever the file gets too large or OH restarts, the logs roll off. The most recent 9 files are kept. The first part of each log statement is the date and time.
Well, that points to something going wrong with the file system. Is it always MapDB that complains or is it sometimes a different persistence engine?
It would be odd for an error in MapDB to mess up charts though. MapDB only saves one value per Item, you can’t chart with it. But if the file system if full all your persistence (and anything else that writes to that file system) would be effected.
If you are running openHABian, than most of OH operates out of zram which, because RAM is scarce, isn’t that big.
When persistence stops, run df -h and see if any file systems are full.
Yes, zram is installed. I also run df -h above, you can see it in my first post, there is a lot free.
I will try this today evening.
My OH3 is installed on RPi 4 with 8GB of RAM, I checked the free RAM before I restarted the Service, there was about 6GB free…
It seems, when MapDB crash, all persistence of OH went down, but the DB itself is still running.
Where can I see where this MapDB is writing the data to?
Memory Settings & Limits for ZRAM are in /etc/ztab → It has nothing to do with the 6GB.
But first check ‘zramctl --output-all’ if any limit is hit, as mentioned by mstormi
I didnt restarted openhab service yet, I will uninstall MapDB, as I’m not using it and then try it again…
I’m not sure, but is MapDB not responsible for the last state of some items, like switches?
Depends on how you have it configured. By default rrd4j will save all Items it supports and restoreOnStartup all of those Items. At this point most Items except Strings and Items with compount states like Color and Location are not saved. MapDB by default will save all Items regardless of type and restoreOnStartup. But InfluxDB also will do the same.
If you’ve customized the behavior using .persist files, than see those files to see what is saved how often and restoreOnStartup.
Looking at the screen shots (which are almost impossible to read on a phone unfortunately) it’s weird that one of your CPU cores is pegged and the other three are basically idle. That points to a process run amok.
