I have node-red pull a backup nightly so I am not too concern about corruption.
Been lurking here, but got nothing definite to add so far. Here are some tangent points:
- C4 free memory report indeed includes about half a G of swap space.
- Running out of memory will freeze the hub. C4s would fare worse because they will try to use (very slow) swap space just before running out of memory. On C7s, Linux's OOM killer can just get rid of the hub process, but C4s are forced to suffer instead.
- There are still places to minimize memory use/prevent leaks within hub code. I'm getting some very useful feedback on this, but changes will have to wait until 2.2.4. The lid on 2.2.3 is welded shut.
- Soft reset tends to cure database related slowness. Based on community posts and support tickets, I'm inclined to build an equivalent of soft reset upon every hub restart. Will let you know when that happens.
11 Likes
Same here, the Hub has slowed and even events are slightly delayed, rolled back to 2.2.2.129 and all back to normal.
I have noticed this consistently when I am having issues that a soft reset resolves the problem. But then issues come back at some point, could be weeks even. But 2.2.3 has defiantly been worse for me then previous releases, I think there were just so many changes in that release that issues are bond to crop up.
1 Like
I did the same, rolled back to 2.129 and everything is very snappy and running smoothly.
I reverted back to 2.2.129 last night. what a difference in overall speed. interface and automations. I even started a thread asking about it a few weeks back and most said it was fine. something is going on, hopefully 2.2.4 irons it out for the C-4.
I have two hubs a C4 and a C5. When I update I usually do so at the same time and the C4 i had been noticeably faster, despite having the bulk of my dangerous apps and drivers.
Since the latest update, the C4 is noticeably slower. I have gone back to rebooting every second day.
Hub just took a dumped again. Couldn't access the main or diagnostic page at all. Power cycled and all is good now.
Well, that didn't last long. My hub is unreachable this morning.
Woke up this morning with a message saying my Zigbee radio is offline. None of the zigbee devices were working.
Try disabling/enabling the zigbee radio. If that doesn't bring it back online, do a reboot.
Is the hub well ventilated and not getting too hot.
Yep that happened to me too.. I was able to stop/start the zigbee network but then rebooted just in case. Has been working okay for the last week. Dont know what happened.
zigbee radio going offline is typically an indicator of low memory state, at least on the c-3 to c-5 hubs.
Hub is sitting on it side vertically and has a 120mm fan blowing on it. Definably not a heat issue. I just preformed a hub reboot to get things going again.
My hub is a C4. Anyone setup node red to retrieve the free memory from the hub for tracking over time yet?
Just thought I'd ask as people have mentioned that if they get too hot it starts to shut down cores (I think that is what was said) and the first thing to suffer is the zigbee radio.
I know when my C3 gets very busy it turns off the zigbee radio. It used to happen 50% of the time when doing a backup. Since the event optimisation things are now good.
The C-4 is my "main" hub - All my rules are on Node-RED. I only have Maker/Groups&Scenes/Lock Code Manager/Lutron/Zone Motion apps installed.
My suspicion is yet another flaky Zigbee device + maybe something with the latest firmware update. Am in the process of moving things over to the C-7 but am thinking I might just use a C-5 for my Zigbee devices.
Yes I added this to my performance monitor flow where I also capture memory on each of my 4 production hubs. As you can see I have had to reboot the hub especially the coordinator a few times in the last week because of slowness. Coordinator is a C5 and rest are C4s. All mounted vertically on the wall.
3 Likes
You mind sharing your node-red flow for the memory monitor?
2 Likes
