Is there a way to write a rule to have the hub notify you automatically if the memory is getting low? If so how? How low would the memory get?
Also, is there a way to write a rule to have the hub reboot itself?
Install @thebearmay 's hubitat information driver.
4 Likes
Is this in HPM? Sorry not where I can look right now.
Also what is a good threshold to say low memory?
It's in HPM - look for Hub Information Driver v3 . As to where to set the threshold really depends on your hub's load and how rapidly you are losing memory. I'd start with where @rlithgow1 has it set in the rule above (which also has the reboot command in it).
3 Likes
I'm fine with 120k but my hub has run as low as 80k
2 Likes
How do I know what my memory levels are? I am learning this side now. It is an adventure. Lol
Thank you I will look it up tonight.
Great thank you. I will check it it tonight.
Oh by the way I absolutely love the name of your hub. 
1 Like
How are you creating that graph?
I have hub info feeding into prometheus, but it would be nicer to have the graphs right in Hubitat.
I know there were some memory bugs some versions back where it was necessary to reboot, etc. I created some routines to monitor/warn/reboot as needed at that time--but I'd forgotten all about them because it's been so long since they triggered.
That makes me think that the hub itself, at least, has memory management well under control right now.
Memory works in conjunction with other factors. A loaded HE with tons of Apps can use more memory while a lightly used device may use very little - it's dynamic. There is no 'if memory drops to 100k I must be in trouble'. I myself set memory min to 120k. I've gone as low as 48k when I was tossing apps into my HE willy-nilly. When I cleaned it up my headroom increased substantially. Think Paging File. Monitor your memory with graph for a week or more - then decide for your self what a safe level and unsafe level breakpoint are.
Buillt in webcore (I don't use webcore itself, just the graphs which basically are Hubigraphs)
1 Like
I didn't get to try the other app mentioned above (I will as soon as possible) but I did notice this morning that my Zigbee Map App that I downloaded from HPM a while back to look at my zigbee mesh shows a memory and CPU reading. The posts above reference 120K and numbers like that. I don't know if it is a difference in apps and how it shows the numbers but mine is showing 1.12 million instead of in the hundred thousand. I have attached a screenshot of what I see. Does this make sense to anyone? Also what is the CPU? I have a C8 Pro if that makes any difference.
C8 pro has twice the memory that's why it's in the millions
1 Like
That makes sense. 
. So sounds like I am ok. I got a little way to go.
What is the CPU reading?
CPU is the load reading (the number of processes that are using or waiting to use the CPU) - a load of 1.0 means that there are enough processes running or waiting to fully occupy one core on the CPU; note that the load can exceed 1.0 if more threads are waiting than can run concurrently. Also need to understand that the hub has 4 cores.
1 Like
Thank you. So basically with 4 cores if they were stacked that target line would be at 4 is that correct? So where I am not even going over the original 1 that means all is good. Am I understanding that correctly?
1 Like
Batting 1.000.
1 Like
Thank you. You guys teach me so much every day.
1 Like
