Raspy 4 with sd card - good enough for this integration? #191
Replies: 3 comments 3 replies
-
hi, I am using an RPi4 with an SSD for 2 years now. Sometines it is a bit slow, but only, if I use the HA Energy manager and want to view accumulated data of a complete year. The performance in general is enough for me (using several HA addons). When it comes to speech recognition, it is not suitable. Maybe I will update to a RPi5 in a couple of month. The best argument pro RPI in general is, that it is so well supported. I never had some hardware / firmware / driver trouble. Regarding this integration: I guess changing the intervals won't affect the endurability very much. SD cards are not really meant for continuous data writing. Slowing down the templates would also result in some wierd behaviour (e.g. the binary sensors "PV producing, exporting... will get delayed) You basically already figured out all the reasons, why I am using an SSD and not a micro SD. The good thing is that they are around 30€ at the moment for 512GB :) |
Beta Was this translation helpful? Give feedback.
-
Maybe Mini-PC between 150-200 Euro is better than RPI and more promising ... ;)) My is a Beelink Mini S12 Mini PC, Intel 12. Prozessor N100 (4C/4T, bis zu 3.4Ghz), 16G DDR4 1TB PCIe SSD Mini Computer W11 Pro, LAN 1000M/25W TDP/WiFi 6/BT5.2 für Zuhause/Büro and it works great under GNU/Linux (Proxmox ==> VM HAOS) with 6-10 Watt (no Link, because I am not sure it is allowed). |
Beta Was this translation helpful? Give feedback.
-
Thank you @mkaiser, @Louisbertelsmann and @mjunker61 for your support. I've switched to an external SATA SSD on my Raspi 4. That cost me EUR 22.40 For others that might be interested: If this will still not fit my needs, I will buy a Mini PC like Beelink oder Intel NUC |
Beta Was this translation helpful? Give feedback.
-
I'm interested on which hardware you run home assistant and if you have also experienced similar behaviour / issues as I experience.
I use Home Assistant OS on a Raspberry Pi 4 with 4GB Ram. I have several add-ons installed like AdGuard, SSH Terminal, Samba, File Editor, Network UPS, Studio Code Server.
Since July I have this integration running and shredded two sd cards already (SanDisk Extreme PRO microSD 64gb, Kingston High Endurance, microSDHC, 32 GB, U1, UHS-I). With lots of entities in this integration (scan_intervall: 10s) and many template sensors constantly updating, there is a lot going on in my system. The database went up from 150MB to around 900MB. A lot of sensor data is written to long-term statistics by default, which will lead to a slowly but surely constantly growing database . I use recorder default purge setting of 10 days, but have set commit_interval to 30 seconds as recommended in the documentation if one is using a sd card.
When Home Assistant is up and running for some days, it starts getting slow when interacting with the UI
What's your experience on this and what kind of setup do you have? Do you think using an SD card with this integration should be no issue at all? And if you use a sd card, do you face similar issues?
I've read different things re using a Pi and sd card:
PS: I've set the scan_intervall from 10s to 20s a week ago to check how much this reduces disk I/O
PPS: I've read that template sensors - depending on what their purpose is, on which sensors they depend and what they calculate - can constantly update (e.g. <10 seconds), which leads to much more r/w operations on the sd card. To address this, I've added the following to modbus_sungrow.yaml:
Beta Was this translation helpful? Give feedback.
All reactions