restoreOnStartup randomly skips Items

[rlkoshak (Rich Koshak)]

@Kim_Andersen

Hmm I thought influxDB couldnt be used to restore on startup…I dont know why though. Maybe its because usually its MapDB which is mentioned in regards to restore on startup.

InfluxDB absolutely can be used for restoreOnStartup. But I do not recommend it because:

• it’s going to be slower than an embedded database like MapDB
• there are boot dependencies, when the machine reboots InfluxDB has to be fully started and accepting connections before openHAB tries to connect to it for restoreOnStartup
• it requires you to use up lots of unnecessary space saving all the changes to Items where all you care about is the most recent change (this can be mitigated a little bit with retention policies).

For restoreOnStartup, MapDB is the superior choice over any external database and it’s superior to rrd4j because it supports all Item types, not just number type Items.

@papaPhiL

all values are in the mapDB at least once

I know you probably don’t mean it this way, but just in case anywone is confuised, MapDB only stores one value for each Item, the most recently saved value.

@TRS-80

I am about to do my own persistence set up, and I think I will do something along these lines (MapDB + Graphite / Carbon in my case, is what I am thinking).

I don’t want to get too far off topic but see Design Pattern: Group Based Persistence and see the list of addons for the list of supported databases openHAB can read/write to. Graphite is not among them so you will probably have to write a new add-on to support it. For time series type databases, most OH users use InfluxDB.