Hi
In past two weeks or so my hub makes randomly disconnects and then it tries to connect again, sometimes it comes online again, but sometimes it won’t, so i have to unplug the hub and replug it and it comes back online (rebooting HUB from IDE is not accessible during this)
Conditions:
HUB firmware 44.00009
For the last month i changed most of my connected devices to beta edge drivers.
Earlier this month i changed zigbee channel from 23 to 20 because of some pairing problems with IKEA 10W LED driver would’t pick up the edge driver.
HUB has been working flawlessly for past three years.
New incident on my HUB
As of writing this my HUB has just been trough 13 disconnections in 35 minutes and then came online for 7 minutes and now disconnect loop again
What is hapening?
Is something on my mesh network is because of Samsung experiments on something before the great transistion. All their systems seems to work
I’m guessing v3 hub since you said you’ve had it 3 years. How many running drivers?
My experience with this from having logcat up while it happens, documented in part in the topic below, is that the hub is running out of memory and processing is slowing to a crawl. Eventually driver events begin to process very rapidly but then slow down again.
This makes sense. I just had problems with low memory and now that you mentioned this I can see these two things having connection. My hub could easily be “full”.
The hub is hard wired to a Netgear ORBI satellite that hard wired to the Netgear ORBI router
Is there a cure for this?
It’s a v3 samsung hub
21 Edge drivers installed from 13 different channels.
I have a total of 188 devices (that includes virtual switches)
14 custom DTH left that can’t survive the transistion.
Yeah this makes sense, is it maybe too many edge drivers installed or will the transistion begining saturday maybe free up some memory in the hub because of the DTH and Smartapps not supported ?
I have around 125 devices on a v3 hub. I’m not sure how much load each device adds versus extra drivers, or whether the efficiency of the driver matters much.
On FW 44.x, my hub was completely unstable with 29 running drivers, going into that failure state 2+ times per day. I reduced to between 19 and 23 and it was stable.
I’m on the firmware beta, and FW 45.x adds an attempt to proactively free up memory by restarting drivers when memory in use exceeds some threshold. This is only visible in logcat, but could lead to timer resets or missed events. With 17+ drivers I saw this happening routinely. I currently have 16 running and seem to be stable, but will likely reduce further for breathing room.
So there are definitely limits to what the v3 can handle. I haven’t seen any official public guidance from ST though so it’s all guesswork to figure out when you’ve crossed the line. The v2 hub has twice the memory, and from what I’ve seen others post may actually be able to handle 50 running drivers.
Remove some drivers, consolidating where you can to have a single driver run more devices. Buy another hub and distribute load between them (I’m not crazy about this idea - both the v2 and v3 are getting to be old models).
Thanks for the info. Because of these limitations, 200 devices, 50 drivers etc, I’ve been thinking about using Home Assistant to relief some of the burden.
Same. 15 drivers on the v3 isn’t enough. It’ll be even worse for non-developers considering that manufacturers like Aeotec are releasing individual drivers for each of their devices. If there were a v4 hub on the horizon that was designed to support local execution then I would go that route.
Here we go again
Just as my HUB updated to 45.00009 yesterday
Disconnects started again
My past solution of uninstalling unused egde drivers and consolidate drivers was a solid solution, because i have not changed anything else since last incident than HUB is updated.
Disconnects seems to be random as before, but with longer interval, but shorter period
Was it just a few disconnects around the time of the firmware update, or are they ongoing? If you haven’t already, I would power cycle the hub so it’s starting fresh. You should also post about your experience in the firmware update topic since ST staff are more likely to be monitoring that.
The 45.x firmware institutes a new (lower) cap on memory usage by edge drivers, and restarts one driver per minute when you exceed that cap. It probably helps if a driver has a memory leak or randomly spikes in memory usage, but I’m not convinced it helps in cases like yours where the baseline is at that cap. Meanwhile, the cap is lower, so you’re likely to start having problems with fewer drivers/devices than what was stable on fw 44.x.
With the new firmware, you should now see a memory warning or hard stop message if you try to add a z-wave or zigbee device by brand. To my knowledge, it’s the only way to check how close you are to the cap. Below is the warning. My hub is generally running ok at this level, though I occasionally see drivers restart. If you get the hard stop message saying that your hub is full then you’re probably still in the danger zone and need to lighten the load on the hub. I think 10-15 running drivers is a safe level with firmware 45.x, but the number and mix of devices probably makes a difference.
Hi @philh30
It is ongoing, but not as often as before firmware update. After a power cycle HUB is stable for a while. As of writing this The HUB has just disconnected for tree minutes since last incident for 14 hours ago.
Maybe when all leftover old smartapps and DTH’s stops working in the near future, that will free up some memory. I have still about 12 smartapps and a few DTH’s. Maybe I should just quit those smartapps anyway
I think if the cab is lower I will test by trying to remove some devices with corresponding egde drivers to see if it gives more stabillity
And I haven’t seen that memory warning yet
Hub has been in a disconnected state since the upgrade. power down and then back on again maybe reconnects everything for a couple minutes, but then right back to where it was. Any ideas? Firmware v 000.045.00009
power down and then back on again maybe reconnects everything for a couple minutes, but then right back to where it was. Any ideas? Firmware v 000.045.00009