I have been noticing that I'm getting about 15 days out of my primary hub before I have to restart due to used up memory. I haven't dug into what is causing this but I feel like it's pretty normal.
Secondary hub could likely go double that but I always restart together.
What's everyone else's uptime like?
2 weeks is more than I have been getting lately.... more likely 7-10 days.
My hub runs continuously unless I perform an update.
I should add I am probably too trigger happy when it comes to restarting.... I have an old C-4 that has a habit of getting pretty high temps and my main rules / local C-7 or my lighting C-8 I tend to restart at a moments notice of memory dipping below 280Mb. That doesn't mean I have seen any degradation in performance... So me saying 2 weeks is a long time... I'm interested now to only keep an eye on temps and let them run right down until I see a performance impact....
Same here. Usually weeks to months, til I trigger an update.
My hubs will run until there is an update and then I was a week or so before apply the update for my personal hubs. Many in the field have been up and running 6 months or more.
That's the same for me.
I have two C-7s tied with Hub mesh: one with 140 Zigbee/25 Z-wave devices, one for network integrations.
If I turn off backups, my hubs run until I do a manual update. If I do backups every other day, I get 2 weeks or so. Then memory starts to fall below 100mb and that triggers a rule I created to reboot.
I use Nodered to trigger daily backups and DL them to my mac… my hub stays online w/o issue untill i update it.
You have me wondering now if backups actually are the source of my memory dropping.
I just checked and I have daily backups set on both hubs. Granted, the primary runs a lot more stuff but it will drop to below 100mb in 15 days where the secondary one gets to about 250 I think in the same amount of time.
I see free memory a lot lower than I used to. Even after an update and reboot, within a day free memory is below 300,000. However the ongoing memory drop-off has been slower.
On 2.3.5.131, I left the hub a full month before updating to .146 3 days ago. It was around 180,000 when I did. 3 days on, it's already down to 236,000. It's difficult to know whether I'll get anywhere near a month before I reboot again (I won't let it go below about 180,000)
This is a reasonable practice for the C-5 and C-7. The C-8 uses a different JVM, and is stable at much lower levels of free memory (like 130-140,000). So my C-8 hubs are set for a lower limit (130K) than my C-5 and C-7 hubs.
That's good to know - I'm on a C8 so will keep that in mind
Back in the early days I had to reboot my hub because it would slow to a crawl and if I didn't reboot it would crash within a day, but that was several years ago. I would constantly watch the memory drop and would preemptively reboot. I don't do that anymore. Even though the free memory space still drops, I just let the hub do it's thing. I only reboot when doing updates. I know I've gone several months between updates so I'm satisfied not worrying about it anymore.
Since there are times when the Hub isn't busy (e.g. Sunday at 4:00am), I now have a weekly reboot.
I shouldn't have to do this, but it's not a great inconvenience.
As well, I also have a rule which reboots when the free memory goes below 180K.
Again, I shouldn't have to do this - I assume that the Hubitat staff are working on this issue, and will get it resolved (sooner or later)!
It's interesting to see that the LAN hub has higher load and higher temp but more memory.
I really wish I could locate where this "bleed" is on the main hub.
I see the same. Also, my hub with most of the devices on it is always the hub with the lowest free memory and the one with the fastest memory drop of the 3 hubs I am running. I have been assuming it takes more to handle devices.
I thought it was just nature of the beast but then I found that others are letting things go for much longer than the 15 days that I get.
Looks like my longest one was back last august where I went what looks like 26 days. The thing is lately though so much has been happening it isn't likely to go longer then about 6-7 days before some kind of update dictates a reboot.
I would let it go until allot longer if the updates were so quick.
Are they both C-8 hubs? Or is House C-8 and LAN a C-5/C-7?
C-8 hubs use a different JVM from older models that consumes more memory when it initially loads.
Download the Hubitat app
