Closed point-4ward closed 6 years ago
@mf-social does the issue go away if you roll back to 0.68? That would be a clear indication its caused by the release.
Ah, now you see - this is the problem with trying to recreate it.
Like I came home last night from work and the powered node was showing dead again, so I restarted the network and it was fine. It's been fine all night. This morning I've restarted homeassistant (0.69.1) 4 times and it's been fine on every restart.
So if I go back to 0.68 or 0.67 and restart 10 times and it always works, there's nothing to say it wouldn't have worked those 10 times on 0.69.1 and that the 11th time it might have come up dead on all three versions, and I don't have the capacity to keep swapping versions for a day here and a day there with random restarts atm. (otherwise genuinely I would, because I need to get to the bottom of this myself, and I would like to get as much information as possible).
Does it sound usual that it would mark a node as dead in less than a second though? I know that a second in computing terms is like a 1000 years, but you would think it would try over a few seconds at least, then maybe try again over a few seconds a short time later before marking it as dead??
(also, thanks to @MartinHjelmare for putting the logs in a hidden section, couldn't work out how to do that!! - cheers 👍 )
Home Assistant release with the issue: 0.69.1
Last working Home Assistant release (if known): 0.68.x
Operating environment (Hass.io/Docker/Windows/etc.): Debian / venv
Component/platform: Zwave
Description of problem:
I've held off on putting this in because I'm struggling to recreate it, and I can't even put my finger on exactly what is up, but at some point over the last couple of releases something weird is going on with zwave.
Basically, I keep getting nodes reporting as dead. I haven't changed anything at all on my zwave network in months so it's unlikely configuration based, and it's not like my walls have moved so it's unlikely that it's an issue with range/whatever.
It mostly tends to be on restart of homeassistant, in that when homeassistant is back up the node shows dead. It has occasionally happened during the day when homeassistant has just been running as normal. Restarting the zwave network (by stopping it and starting it again a few seconds later in the configuration panel) usually brings it straight back up.
Note that I have one powered and one battery node, so it recognises that the battery one may be asleep and doesn't mark it as dead anywhere near as often, but the powered node (which is the one I really need to work, typically!) reports dead on a maybe 8/10 restarts.
It seems as though it's not giving the nodes enough time to reply, and then instantly marking them as dead. Once marked as dead there seems no way to bring them back other than restarting the network. I know one of the options has the text "this may be able to bring back dead nodes", but I don't think this is correct as if you use that option and then view the OZW log it sends a message, and then less than a second later cancels it as 'node is presumed dead'. (This issue isn't new, I came across it ages ago when I had something else wrong).
I've included a stacktrace below that is the complete log after a reboot (whole device) at 11.19 hours this morning. The node reported as dead, and at 11.25 I stopped and restarted the zwave network. The log entries all appear to be entity_registry related as opposed to anything actually zwave network related, but there it is. By the end of that little flurry in the homeassistant log we were back in business,
The OZW log for the same period is ridiculously long, but I've posted it for completeness and basically does exactly as I mentioned above. The system boots, it sends out a message, doesn't get a reply from the light switch and marks it as dead straight away.
Then when I restart the network, lo-and-behold, everything is tickety-boo. Interesting to note though that sometimes when I've had a dead node and I restart the network, the node that was previously showing as dead shows up as 'unknown', but still works fine. Sometimes it just comes back as 'Ready' like normal. I mention this specifically because today is one of the times it has come back as 'unknown', so there might be something in the logs that gives a better clue this time round?? I can't see anything, but....
Additional information
Pfft - no idea - I've tried to recreate it successfully, tried hard reboots, soft reboots, homeassistant restarts, heal networks, test networks.... Dunno, but it is definitely getting worse, and definitely only since upgrade to either 0.69 or 0.69.1 (I can't remember how long I've been trying to work this out to make this bug report better - too much going on IRL).
I'm hard pressed to believe that both the nodes are suddenly failing at the exact time I upgraded homeassistant, and they seem to be working fine when I do get them back up in homeassistant, but I have no other way to control them anymore to check anything else out.
@andrey-git @perosb - you guys seem to be the zwave oracles - any thoughts?
Sorry for the below, but, you know, thorough...
Problem-relevant
configuration.yaml
entries and (fill out even if it seems unimportant):Stacktrace
OZW log