Yeah just realized I forgot to rename the nvram backup file. Currently when executing znp_nvram_reset, it calls znp_nvram_backup and it generates the file as nvram_backup_date_time.json. Next I’ll try just a normal nvram backup/reset/restore to see if that works
2022-01-08 15:34:54 INFO (MainThread) [custom_components.zha_custom] Running custom service: <ServiceCall zha_custom.execute (c:e4cba7611269e3c79cd451767b80a50e): command=znp_nvram_restore>
2022-01-08 15:34:54 DEBUG (MainThread) [custom_components.zha_custom] module is <module ‘custom_components.zha_custom’ from ‘/config/custom_components/zha_custom/init.py’>
2022-01-08 15:34:54 INFO (MainThread) [custom_components.zha_custom.znp] Reading NVRAM from device
2022-01-08 15:34:56 WARNING (MainThread) [homeassistant.components.zha.core.channels.base] [0x7EC9:1:0x0702]: async_initialize: all attempts have failed: [DeliveryError(‘Request failed after 5 attempts: <Status.NWK_INVALID_REQUEST: 194>’), DeliveryError(‘Request failed after 5 attempts: <Status.NWK_INVALID_REQUEST: 194>’), DeliveryError(‘Request failed after 5 attempts: <Status.NWK_INVALID_REQUEST: 194>’), DeliveryError(‘Request failed after 5 attempts: <Status.NWK_INVALID_REQUEST: 194>’)]
2022-01-08 15:35:08 INFO (MainThread) [custom_components.zha_custom.znp] Saving NVRAM to ‘/config/custom_components/zha_custom/local/nvram_backup_20220108_153454.json’
2022-01-08 15:35:09 INFO (MainThread) [custom_components.zha_custom.znp] NVRAM backup saved to ‘/config/custom_components/zha_custom/local/nvram_backup_20220108_153454.json’
2022-01-08 15:35:09 INFO (MainThread) [custom_components.zha_custom.znp] Restoring NVRAM from ‘/config/custom_components/zha_custom/local/nvram_backup.json’
2022-01-08 15:35:09 ERROR (MainThread) [homeassistant.core] Error executing service: <ServiceCall zha_custom.execute (c:e4cba7611269e3c79cd451767b80a50e): command=znp_nvram_restore>
Traceback (most recent call last):
File “/usr/src/homeassistant/homeassistant/core.py”, line 1511, in catch_exceptions
await coro_or_task
File “/usr/src/homeassistant/homeassistant/core.py”, line 1530, in _execute_service
await handler.job.target(service_call)
File “/config/custom_components/zha_custom/init.py”, line 73, in custom_service
await handler(
File “/config/custom_components/zha_custom/znp.py”, line 180, in znp_nvram_restore
nvram_obj = json.load(f)
File “/usr/local/lib/python3.9/json/init.py”, line 293, in load
return loads(fp.read(),
io.UnsupportedOperation: not readable
@le_top , latest update has messed up the files, can you please have a look. zha_custom folder now contains two folders another zha_custom folder and local folder.
When performing a znp_nvram_reset, the backup is made for safety and a timestamp is added to avoid overwriting a backup that is actually restored from.
The znp_nvram_backup will backup to the nvram_backup.json file directly (if command_data is empty or not set).
I manage two setups, on one of them I installed zha_custom through HACS.
In case zha_custom is in the custom_components folder, then the zha_custom/zha_custom subfolder will not be of any effect and you can delete that - I do no know why that appeared. The ‘local’ folder is where the output files are written and it should be maintained across updates.
Ok, that should be fine.
You can install using HACS - I’ve submitted a request to add zha_custom as a default repository, but there is some delay. In the mean time you can add this as a custom component by providing the github repository url directly.
@le_top ,Thats exactly what I did, I installed via HACS, I just dont know what went wrong during the update. even my backup was deleted automatically . but everything is fine now because I have made another backup now and it went smoothly .
Yeah I just don’t seem to have any luck with my devices rejoining after doing an NVRAM restore. The nvram backup seems to generates fine in both the backup and reset methods however. In any case, since the high level backup works I think that should be sufficient for most users.
c4664700 (lsb first) → 0x004766C4 (hex, MSB first) → 4679364 (base 10) = the FrameCounter in the nwk_backup.json .
So maybe increasing that number (first 4 bytes of the field above) helps. Adding 2500 is adding 0x0800 or 0x1000 to make it simple in hex and keeping close to that increment.
So “c4664700” would become “c46e4700” (+0x800) or “c4764700” (0x1000), the latter probably being the easiest in most cases. If that’s the trick, then this could be “automated” as well.
But as the more generic restore is functionnal, we can very well skip this.
Gotcha, unfortunately my HA instance crapped out on me so I ended up migrating everything to z2m. I don’t have another TI based coordinator so I can’t test znp commands anymore but I’m glad it’s in a state where the high level restore can be done without too many issues.
Hi
I have 30 devices in zha, but when I open the backup JSON file there is consistently 18 devices backed up.
9 of the devices missing are hive mains powered bulbs and a sonoff motion sensor.
It managed to back up the other sonoff motion sensors.
Any reason why there isn’t the 30 devices backed up?
At least we have this in common - I only have 8 devices in my backup and there are 18 in my network and I have only 2 listed as children.
According to the documentation of the “open coordinator format” this list concerns the children and devices with APS link keys shared with the coordinator. I suppose that the other devices are linked to their routers.
The backup procedure and format is implemented in zigpy-znp - zha_custom is merely reuses/adapts the cli interfacecode to provide it as a service.
I am willing to add new items to zha_custom if I do not need to spend too much time porting them. Time is my main limitation.
For instance, adding the flashing procedures already provided in Zigpy-* is fairly straightforward.
Adding JemerT’s project is too much work:
it’s not provided as a library (no pip install AFAICS);
If there is a unintegrated patch, it’s likely not part of the pip install (but I think one can pip install from a repository if it supports pip install);
Testing. Firmware upgrade is a critical procedure and takes time to test.
Stopping (disabling) and restaring ZHA could probably be added to the service during the upgrade procedure. Data backup (for safety) and restore (if needed) could also be added to the procedure.
I do not nkow to what extend the Python Scripts Integration gives the scripts access to the serial ports/sockets.
That integration could be a first step to porting to zha_custom by testing the firmware upgrades scripts as they are in the HA environment.
I suppose that zha_custom could probably call another python script natively.
The external scripts can possibly also be fetched using the exposed zip deliveries in the repositories.
Ultimately - if the respective licences allow it - the firmware binary could also be preloaded or fetched from an external source (listed in some file in zha_custom so that an upgrade to “latest” is possible).
But it’s mainly my personal “human” bandwidth problem. I can help out with blocking issues and integration, but not do all the development.
I’ll update zha_custom when that is available - I already noticed something was in the making.
Backing up was not my initial reason for extending the original zha_custom ! But I did add a daily coordinator backup to my system.
In UI mode, all commands are available in the drop-down and you can select a ZHA Entity from the second drop down.
For quite a few commands that will do the job.
If you need more than just text for ‘command_data’, you can still select the command and the entity in UI mode and then switch to YAML mode to finish setting up the service call.
You can still enter the IEEE address in Yaml mode if you want to.
Ok, this is something that will be useful to those that want to avoiding reading zigbee attributes from the database.
The zigbee attributes you read or write (and read back) can now be written directly to home assistant states or state attributes
For instance, I want to read apparent power from my electrical meter and it’s not available through the standard functionnallity, I can now call this service repeatedly:
That reads the attribute and on success updates the attribute. Without “state_attr”, it updates the attribute itself (and you’ll get the updated time as well).
