Hi!
I’m already seeing 20% lifetime used on the eMMC of my Home Assistant Green. Should I take any actions? Is it possible to pop in an SD-card to move some data over there or what?
This has suggestions on reducing I/O:
2 Likes
Thanks, I’ll have a look. Too bad that it’s relatively complicated to fix this on a product that’s otherwise pretty much ready to use OOB
So if I add an SSD to a USB port on the HA Green (after buying a powered USB hub…) would that mean that I wouldn’t have to mess with the database?
I don’t like that I have to add more devices to the Green, but if that can save it from becoming e-waste, I guess I’ll have to.
I have a bunch of 120GB SSD’s here in good condition.
Do I have to manually unplug and copy out backups from that SSD to keep the database if the SSD should fail though?
My database is 710MB. I’ve checked with SQLite now, and it doesn’t really seem to be anything out of the ordinary. For events, 50% was recorder 5 min updates.
But I did find this:
Everything seems to be at 1000-5000 cnt, except these that are much higher.
Should I disable logging for all of these? Won’t that mess up graphs for power consumption etc?
Should I also do something about the “last seen” entries?
I don’t need a history of “last seen” and my logbook on Home Assistant is filled with them. It’s nice to have the last value of course, should it be out of battery or similar.
I actually believe the “last seen” entries is because of z-wave devices updating:
So they’ll update at a set interval, or if there’s a certain amount of change. The default was much more often, like ever 60 seconds vs my current 300 seconds. I’m not sure if it makes sense to reduce them even more (they’re not battery devices).
Update: I changed Temperature and Meter Report interval to 900 now.
I like Shelly Wave’s approach better, they have a minimum interval instead.
I also fully disabled Tibber Pulse as it was reporting every second… I don’t know how to make it report less, and I don’t want to ruin my HA Green.
Is this an issue with the Tibber Pulse integration that needs to be fixed?
Search this forum for posts on setting up and using an SSD.
Regardless of how data and devices are managed, have a good, tested, backup / restore strategy. And move the backups off the HA Green.
Integrations have been known to be problematic. As an example, VSCode is notorious for being a resource hog. This link describes how you can profile Integration behavior:
2024.5+: Tracking down instability issues caused by integrations. Any issues with Integrations should be reported on GitHub.
Also check out: The Home Assistant Cookbook - Index.
1 Like
I already have backups set up to copy over to my NAS, so there’s no worries there. I also have both SSDs and USB chassis for it, so that’s no problem.
But it doesn’t seem to be right to have to use an SSD for it, if the eMMC should handle it though.
I contacted support, and they also thought 6 months of use and 20% lifetime was excessive.
In case you weren’t aware (even though support should know this), lifetime used is a rounded up number. Basically, anything between 11% -20% will still be reported as 20%
I see. But that is still a very high number though after just a few months. I haven’t really started using it yet either, just added devices, no automations or scenes are set up yet.
I read that Recorder can be configured to write less to DB, but that it would introduce “lag”. I see that it writes every 5 second as default, so what would happen if I set it to 30 or 60 seconds?
Will it mean that it takes 60 seconds before entities are updated, or dashboard/graphs are updated?
I thought it would just use memory and write less frequently.
If anything I’d like it to write only once every 15 minutes f.ex, or in reboot.
I don’t see the downside with having less writes?
Well…f it… I just jammed a USB SSD in there, and moved disk.
At least now I won’t have to worry about Tibber Pulse or any other integration writing often to eMMC. Without Tibber it seems like I’ll be able to reduce my database to just 50MB. But as I want the power status, I’ll just have to use that integration. I just wish there was a feature to cache f.ex. 1 or 5 minutes and then save, to reduce writing.
But honestly, I’m not really noticing any differences between having a 50MB vs 800MB database though, it seems to be as snappy as when I first installed HA 