Aeon HEM losing children

[rearden]

I have a zwave Aeon HEM, veraplus ui7.  It created three devices, the total power and then one for each of the two sensors.  I have it installed in my panel to measure my home's power usage.  After a few days the children stop being updated (the separate values), but the parent (total power) is still updated on a regular basis.    I think it is a Vera issue because the parent updates fine, the data is coming through, but the child devices stop. I have contacted Vera support and they cannot solve the problem.  One fix was to recreate the devices every 24 hours; that had the side effect of incrementing device ids and therefore all scenes and Power Arithmetic break every 24 hours.

Anyone have any ideas on why this is happening?  or a possible solution?

rearden

[fitz2380]

Do a search in the forums and you will find several posting related to this.  I have posted several myself and have never solved the solution yet.  Several people have claimed to, but anybody I had contacted related to this still was having this very issue. My AEON HEM are GEN1.  I don't know if this issue occurs with the GEN5 version. 

[reneboer]

Hi,

I have the same issues with a Gen2 clamp. Typically after a Vera firmware update the child devices stop reporting. I found that doing a Reconfigure node right now on the main device gets things going again. However, the child devices get recreated with default names and different device ID numbers and thus braking scenes, datamine logging etc.

So I made this little script you can use. Configure it once with the correct values for the names and device IDs. Then when they stop again, do the Reconfigure node right now and get the newly assigned device IDs by altids e1,e2,e3. You plug the new device IDs in the script, run it and you are back in business. The script has values for mine in as an example.

Cheers Rene

[fitz2380]

I have learned that it only takes a reboot to lose the HEM children.  This can be detected by looking at the capability strings of the children and when they go null, they are broken.  I have captured this issue after a reboot and it appears to me it is probably a timing issue.  I thought at one time that restoring from a backup recreated the children, but I have not been successful and repeating this.

I wrote a script to detect the loss of the children and then i turn on a virtual switch when that happens so I get notified.  I wanted to create a script similar to what Rene did so thanks for providing that.

Fitz

[reneboer]

Hi Fitz,

Thanks for the pointer to the Capabilities. I added a notification on that so i can see when they get wiped.

Cheers Rene

[rearden]

Your script may be the bandaid I need!  Is there a way to automatically determine the new device ids after a reconfigure and then I can just run a scene with a reconfigure and reassign ids every week?
rearden

Hi,

I have the same issues with a Gen2 clamp. Typically after a Vera firmware update the child devices stop reporting. I found that doing a Reconfigure node right now on the main device gets things going again. However, the child devices get recreated with default names and different device ID numbers and thus braking scenes, datamine logging etc.

So I made this little script you can use. Configure it once with the correct values for the names and device IDs. Then when they stop again, do the Reconfigure node right now and get the newly assigned device IDs by altids e1,e2,e3. You plug the new device IDs in the script, run it and you are back in business. The script has values for mine in as an example.

Cheers Rene

[reneboer]

Hi,

You can identify the new child IDs. The parent device is not changing so you could loop over all devices, looking for the ones that have the id_parent attribute set to the parent and then use the altids (e1, e2, e3) to map it back to the original data.

Update: I could not resist  Attached is an updated version of the code. Not fully tested though. Waiting for the next time the reporting stops.
You could go even further and look at start up if the child devices IDs changed and the restore them, but I won't go that far yet as it is a tad risky approach with an automated reboot included. You could end up with a Vera stuck in a loop. So sticking with a manual version for now.

Success.

Cheers Rene

[rearden]

Thanks!  I have placed the code in my recreate scene.  I will tell you how it goes.
rearden

[fitz2380]

I wanted to kick this topic to see if there has been any progress or if anyone's HEM meters continue to work over time.  I have given up on creating any scenes for the children meters as they usually get messed up within a week or so after recreating them.  The main meter device does continue to work and update correctly but as noted here and other posts, after a restart at some point the children get munged and do not work. 

Anyone having success in keeping these alive or maybe everyone has given up on these.  The previous posts provided ways to detect the munged children, but I do not know of any way of restoring them as I have never gotten the restore from backup to work again on restoring the children.

Fitz

[reneboer]

Hi,

For me it has been running fine since the last reset using the script above back in Feb. I did two firmware upgrade since that did not have any negative impact.

Cheers Rene

[fitz2380]

Just so I understand, you have not had the child devices go bad on you since February?

[reneboer]

Just so I understand, you have not had the child devices go bad on you since February?

Correct. All nice and dandy. This does happen in a Vera environment from time to time

Cheers Rene

[rafale77]

I have discovered why this happens to the vera... as I have 2 Zwave controllers and more network and this problem does not happen to home assistant, it is not a problem with the HEM or the zwave network. Turns out, it is a variable corruption of the child device which likely occurred during a luup reload. Decreasing luup reload frequency helps but I also discovered that recovering from a backup of a time when the child reporting was not lost, also recovers the reporting. I have not yet looked at which variable in the child device is getting corrupted but it is caused by something in the user-data.json

[reneboer]

Hi,

I had a problem with the HEM at one point and could not get it to work at all again. It turned out that one of the three power ground wires had gotten loose. No idea if that was the root cause, but I have not had any issues since then.

Cheers Rene

[fitz2380]

rafale77 - I have determined that the corruption occurs during a luup reload.  I have been able to trace through my logs and can see when the 'variables' of the children get set to null or 0.  This happens during a luup reload either from a sporadic reboot or some other user created change which causes luup to reload. 

It appears that you have been able to restore from backup.  I have not been able to consistently restore from backup.

Given that the failure occurs on a reload, I am wondering if it is a timing issue for the device to respond which corrupts the children.  I have two AEON HEM meters and they both were located in my breaker box about as far away as they could get from my VeraPlus.  I have now moved (1 week ago) my controller to the center of the house to see if this will have any impact on these devices.  I will update when enough time has past.