Slow zwave network after "heal"

Setup, Configuration and Use Runtime
1
  • Platform information:
    • Hardware: Pi 3 - Razberry module
    • OS: rasbian
    • Java Runtime Environment:
    • openHAB version: openHAB 2.5.0~S1603-1 (Build #1603)
  • Issue of the topic: 60 - 120 sec delay after “heal”
  • Please post configurations (if applicable): N/A
    • Items configuration related to the issue: all zwave
    • Sitemap configuration related to the issue: n/a
    • Rules code related to the issue n/a
    • Services configuration related to the issue: n/a

Upgraded from 1525 (ish).
Reboot of the pi fixes the problem, things become responsive again. But after auto “heal” it slowly chokes. Adding of new devices fail. looks like zwave network is swamped.
If applicable I can start debug logging and provoke the issue.
Any quick fixes?

//JAF

2

There is an issue with heals at the moment. I made a change on Saturday to try and fix this. I think this should now be included in the latest build, although as builds have been unstable recently and most have failed, I’m not 100% sure.

In any case, please use the latest binding and let me know if that works.

3

Cheers Chris, will do.

//JAF

4

did a upgrade to openHAB 2.5.0~S1606-1 (Build #1606), problem persist. Do you know the build number that are stable from this error?

//JAF

5

Can you provide a debug log please?

6

yes, I started debugging now and changed the heal to run at 3PM GMT+2.
Will upload logs in aprox 45 mins.

//JAF

7

Thanks.

8

@chris I’m wondering about the change you made to wait for a response on the neighbor update. I see that transaction has a 75 second timeout. Will other transactions be queued until the neighbor update completes or times out? I’ve seen successful neighbor updates take 20-40 seconds on my network of about 100 nodes.

Edit: BTW, I just installed the latest build. As soon as the network settles down I’m going to do a heal on one of my battery devices.

9

ohablog2.gz.txt (100.5 KB)
ohablog1.gz.txt (813.8 KB)

Hope i did capture relevant data.

Heal was at 3PM today. extended debug started aprox 2:45PM

//JAF

10

I need to double check. In general though yes, the system will wait for a transaction to complete before starting the next one.

The timeout in ZWave (from memory) is 75 seconds. The controller may terminate this earlier, but I have set the timeout to 75 seconds for this reason.

11

Thanks. It looks to me like the initial issue where we weren’t waiting for the response is solved. It now seems to wait, and it blocks for the full 75 seconds before the next transaction is sent. The fact that it receives the DONE message is ignored and it waits for the timeout which means the HEAL doesn’t complete.

I will take a look at why the DONE message doesn’t complete the transaction tonight.

12

I have upgraded to the Build #1607. It appears to be more responsive now. How ever the battery units are kinda dead in the water after heal. Couldent wait so i rebooted (instacure). I might be able to do more debug logging tonight. It may be unrelated, but i had a hard time including new devices to the network too. Even if most units were responsive.

BR
JAF

13

System appear to be stable again after heal. Even battery devices are ok.
//JAF