Had a right nightmare with HA stability yesterday, added my SMLight zigbee radio the day before and added a few zigbee devices to test setup etc. I then noticed yesterday morning that HA was offline, so logged onto Proxmox to checkout the console and saw an out of memory crash. Bit odd as it's been allocated 4Gb and I could see from the graph it had just spiked.
Anyway it rebooted itself and I noticed a short while later the GUI had lost connection again. Back to the console and could see the out of memory error. I created an uptime sensor (no idea why I didn't have one already) and over the course of the day I could see it kept crashing every 23 minutes.
I went through the process of disabling ZHA as that was the last thing added, but no joy. I looked on the HA forums to see if anyone else was having stability issues with the latest release but nothing similar. Decided at this point I'd bump this install over to a different IP and then perform a VM restore from a couple of days ago to its place.
Gotta add, my first time using HA in a VM on Proxmox (still running my original config on raspberry pi until everything has fully ported over and stable) and the ability to be able to perform a full VM restore and booting HA back to full state all within a matter of minutes is an absolute blessing.
Anyway, very strangely my restore from a few days ago also started hitting out of memory errors despite being adamant this was not an issue at the time of the backup. I should add it's not a host issue either - I did check and I've never exceeded 40% of the total RAM, so host had plenty of memory for the VMs. Decided I'd scratch that install and go back a little over a week - just before I upgraded to the 8.3 release.
Having spent a few hours yesterday re-adding everything I had changed/set-up in the week since that current backup, it is now in a stable state. I haven't yet run through any updates - will modify my backup schedule to include the new VM and once I have a backup to roll back to I'll test the upgrade again.
Bit of a waffle but having in the past gone through the internal backup/restore functionality that HA provides (from what I recall) taking a couple of hours to fully restore, it sure was nice to be able to restore a fully working snapshot in a matter of minutes.
Anyway it rebooted itself and I noticed a short while later the GUI had lost connection again. Back to the console and could see the out of memory error. I created an uptime sensor (no idea why I didn't have one already) and over the course of the day I could see it kept crashing every 23 minutes.
I went through the process of disabling ZHA as that was the last thing added, but no joy. I looked on the HA forums to see if anyone else was having stability issues with the latest release but nothing similar. Decided at this point I'd bump this install over to a different IP and then perform a VM restore from a couple of days ago to its place.
Gotta add, my first time using HA in a VM on Proxmox (still running my original config on raspberry pi until everything has fully ported over and stable) and the ability to be able to perform a full VM restore and booting HA back to full state all within a matter of minutes is an absolute blessing.
Anyway, very strangely my restore from a few days ago also started hitting out of memory errors despite being adamant this was not an issue at the time of the backup. I should add it's not a host issue either - I did check and I've never exceeded 40% of the total RAM, so host had plenty of memory for the VMs. Decided I'd scratch that install and go back a little over a week - just before I upgraded to the 8.3 release.
Having spent a few hours yesterday re-adding everything I had changed/set-up in the week since that current backup, it is now in a stable state. I haven't yet run through any updates - will modify my backup schedule to include the new VM and once I have a backup to roll back to I'll test the upgrade again.
Bit of a waffle but having in the past gone through the internal backup/restore functionality that HA provides (from what I recall) taking a couple of hours to fully restore, it sure was nice to be able to restore a fully working snapshot in a matter of minutes.