We do enjoy our little spats.
Happy to have that discussion about the details if you wish.
Humour me and turn it off and see if makes things worse or better or makes not a blind bit of difference.
A good level of robust debate is a healthy thing 
.
I doubt it will make any difference - I’m not disagreeing with you about that. What I was saying is it does not cause Armageddon - it doesn’t hurt, it might help, and for some devices, it is probably necessary, so why not leave it.
It doesn’t hurt, and it means that if propagation changes, or something in the house changes, then the system sorts itself out. This is my point - we define systems that are usable for everyone - not just people with the latest nodes and latest controllers. Recommendations should be for the masses, and telling people that doing something will destroy their system doesn’t help.
and there is our difference in attitude. You the optimist me the pessimist .
To you all change heal brings is good, to me it could go wrong. Possibly you enjoy this stuff so like to tune and fiddle to me it is something I want to leave alone and let it do its job.
I did write my own controller for kicks when the SDK first became public but it is hard work and takes time. I may do again with Z/IP to preserve my investment, it is just I do not want to. Also as you enjoy it and have done such a good job why would I spend my time doing it. My turn for a 
so back on subject rather than winks and robustness that is fringing on rudeness.
So this fictional user has a couple of fuses flick tonight. A load of his nodes drop out of the routing table or get worse routes due to the heal.
If he had not healed he would have woken up reset his fuses and routing would return to normal as the table would not have changed due to a few failed commands.
So for the masses how do they balance the risk?
Is it your optimist view that all change heal brings will improve the table and they may gain a little on something that already works well
or
my pessimist advice, leave it be and do not risk making things bad if it works very well.
I think now they are free to go either way they can chose mister optimist or mister pessimist.
I’ve had my daily heal disabled for a while now due to the known issues in the snapshots. But I found that there were missed events. I finally got Zniffer setup (the links on Silabs site are broken, but support provided it. I flashed the zniffer firmware to an Aeon Labs gen5 zstick (Aeon Labs support provided the original firmware, so I can flash it back). From initial inspection, the routing looked to be a complete mess, even though no devices had recently been moved or were having issues. Yesterday, I individually healed each mains powered device a few times. Today, the network appears to be performing without missed events. I use an HUSBZB-1 for the controller and my devices are all several years old (~130).
Short version… in my experience, the heal is necessary to keep the network sorted.
@chris, what version of OH and the binding do you use, and do you have your daily heal disabled?
1 Like
My advice totally. Diagnose the issue and fix with individual heal. Now leave it alone.
If you run full network heal a few times my pessimistic nature says you will have issues again.
Absolutely not. I want something that works - I don’t want to “tune and fiddle”. These features don’t tend to do any harm, and they have definitely helped in the past. With the wide range of devices that are out there, in my opinion, on the balance of risks it is better to leave it enabled.
I’ve not really heard of any problems with the heal over the past 6 years that it has been running in the binding (and again, I exclude the present bug that was introduced recently in this!). Most people have it enabled (by default it is enabled, and I’m assuming most people don’t disable it) and we don’t have lots of complaints about it.
What I would expect if people disable the heal is more complaints that the network graph is not working, as the neighbour tables are updated during the heal.
That is an interesting thing I had noticed that the two were tied.
I was going to go and have a look but use zniffer so it does not add a lot.
I also have the extended send data hacked into a version and the extra info that returns to the app is also useful if I did not have zniffer recording and the node supports. Hey it all works so not rushing to change and I am no Java developer or fan of Java.
Hi Scott
Just reread that. That is a very interesting observation.
If you have not run heal the routing is what the routing was when it was last running well. All that can happen is that through a series of fails LWR changes to something and in your case something not reliable.
So if you have not run heal how did the routing get in a mess?
So you have healed the culprits and now the LWR will settle to what works. If it does not you must be getting intermittent failures and LWR is hopping around or even finding brand new route using explorer frames because unless there is a bug in the firmware that is the only time it will change.
Now if you do not run heal again the only way routing will change is if explorer frames find things that are not in the neighbours list. The only other thing that can become LWR is based on the neighbours in the routing table so is not a change beyond the routing table.
Please could you keep your zniffer running and see how things work. It is the sort of thing I have been tracking but unfortunately my production network is behaving too well. It would be interesting what in your network is triggering the new less good LWR
Apologies for losing it there. It is very hot in my office.
1 Like
I’ve no idea - I don’t use versions off the web. It’s from a couple of months back though. Heal is enabled.
It is almost certainly from before the change was made to the neighbour update. Unfortunately until the IDE gets fixed I’m a bit screwed with doing developments which is (to put it mildly) a little frustrating.
3 Likes
What happened with the IDE? Piqued my interest!
From reading this entire thread with the back and forth, I understand some more about how Z-Wave works, and I’m also more confused than ever, and I’m thinking that probably means I’ve understood correctly. 
It still seems like more intelligence with regards to when to heal could be a good idea though.
And, at some point I should turn one of my sticks into a Zniffer.
I have an Aeotec Gen 5 stick with its jurassic never-changing firmware and annoying disco light, could that one be turned into a Zniffer since I will never use it in production?
I have the same thing and I’m interest in doing the same. I would buy another and connect it to a different box to not interfere with OH’s stick.
Best, Jay
Zniffer discussion has moved to…
I think you have understood.
Your home and your setup is unique and what works for you is not always best for another.
The variables go on and on: materials in the building; the physical distances; the number of devices; the mix of devices; the electricity supply: the amount of reporting configured; the ratio of sleeping device and number of devices that have reporting and how often.
Is there even a quiet time on your network that can support the volume of traffic of a full rebuild? If there is not it is not necessarily going to do good things.
My advice is understand the specific issue you have before taking the medicine. If the network is not ill why does it need to be healed?
I don’t know. I believe daily heal is on by default in the binding, I did not specifically choose to turn it on. I could be remembering wrong, it’s been a while since I installed openHAB at this point.
It is so take your medicine as it may cure something you do not have.
Yeah, I believe you’re stretching that analogy too far. Unlike actual medicine, I don’t see how calling heal unnecessarily is going to cause heal-resistant future network problems.
Preventative maintenance is always a good idea. As I’ve repeatedly said, it doesn’t do any harm (at least it hasn’t for the past 6 years!) and it is required for older devices that have no other way to form routes. Therefore, on balance it is more likely to benefit than not.
Cool. I’ll create an issue about stateful heal intervals and hopefully some rainy day in the future after the IDE works again you’ll feel like writing the feature?