[TODO]: Async

[ArnoutD]

Todo

The Async branch is working stable at my end but that is a set of 1 Circle+ and 8 Circle -F types

I must say that I can only tell AFTER the first intial run where the caches have been filled up so you do not get a load of startup messages. I have seen odd energy baviour at a point in time where I flushed the cache and woke up with this pattern:

but after restart it settled fine and keep running

Here are a few of my observations:

1) The sawtooth in the above image was caused by the counter rollover and then missing the slot 4 download, that is gone

2) The energy usage resets bases on server timezone (when my Ubuntu VM is in UTC it rolls on 01:00 in the graphs, CET rolls correctly, as the python-plugwise_usb does no home assistant maybe we should tell it on startup about the HA timezone to correct the last reset time to align in the graphs

vs

3) When a node has nothing attached to monitor it will not show data: Power pulse counter for node 000D6F0000995D02 of value of -1, corrected to 0, when you attach a load everything comes back

4) Strange thing is that the Circle+ ping times are a lot longer (80ms+) then the circles....(20ms)

my branch is till a bit to chatty on energy as I was monitoring the sawtooth but I cannot test any further as I do not have SED or other plugwise equipment.

I do not know how the relation is with my underlying zigbee network, maybe it behaves different in less stable networks. I can up the plugwuse_beta PARALLEL_UPDATES up to 8 without having issues. Also since i implemented one continious loop for reading and queing/writing I not need any sleeps and I do not have a lot of "network" issues in the normal run. When all nodes are queried for a lot of missing addresses there of course more congestion.

[ArnoutD]

also we should rename to

powerusage\<mac> energy_consumptiontoday\<mac>

for backward compatibiltiy

[brefra]

I've just tagged as a prerelease version of the integration: https://github.com/plugwise/plugwise_usb-beta/releases/tag/v0.40.7a1 This version of the integration migrates the current unique ID of the entities, so no need to rename. It is also linked to the uploaded https://test.pypi.org/project/plugwise-usb/0.40.0a3/ backend version.

[ArnoutD]

Installed 0.40.7.a1 in my production box 8+1 nodes took 6 minutes to initialise

all entities migrated correctly indeed, retaing history and settings

during initial energy collection got a few: ` Logger: plugwise_usb.connection.receiver Source: runner.py:188 First occurred: 07:51:04 (16 occurrences) Last logged: 07:52:07

No subscriber to handle CircleEnergyLogsResponse, seq_id=b'A290' from 000D6F00009964FC No subscriber to handle CircleEnergyLogsResponse, seq_id=b'A2F5' from 000D6F000099637F No subscriber to handle CircleEnergyLogsResponse, seq_id=b'A306' from 000D6F0000995F4E No subscriber to handle CirclePowerUsageResponse, seq_id=b'A304' from 000D6F0000995F4E No subscriber to handle CircleEnergyLogsResponse, seq_id=b'A309' from 000D6F000098AB28 `

no 'warnings' after

[ArnoutD]

0.40.7.a1 24h later ~250 No subscriber to handle ... ~112 No NodeFeature.POWER sensor

both unrelated.

All rotate nicely at 00:00 CET, some very low power usage have a small hickup after the reset ( but maybe just waits for a pulse to happen)

this particular one "lost" updating for some time now... will wait if it restores in the end (putting something in the microwave might help)

I think both warning sets are mitigated by my PR with the await instead of gather and the 5 resend instead of 20

[ArnoutD]

installed [v0.40.0a6] through [v0.40.7a2] warnings remain at 26 occurances of 'No subscriber to handle xxx after 10 retries' over various nodes over 24h of runtime

will activate INFO logging to see if there is a "logic" to it

still graph "issue" of the daily reset of low power using nodes, i doubt of it is solvable and due to the PULSES per Wh resolution:

[brefra]

I do see some No subscriber to handle ... too. Haven't found the rootcause of it yet. ~The graph issue seems to be some kind of HA behavior. It's only applicable when the (rounded) energy usage is 0.~, Found issue, it's due to https://github.com/plugwise/python-plugwise-usb/blob/async/plugwise_usb/nodes/helpers/counter.py#L293-L294. Guess we should guard for negative values only. Will check tonight

[ArnoutD]

Got in a intereseting state today: 2024.03. [v0.40.0a6] through [v0.40.7a2] on 2024.3.0

longer time.... something " broke on 13:00 for this node but for other nodes it is different

reset to a1 but no change

my cirle+ is still going strong by my circles are bugging.,.,. I will move my cache and compare

removing the cache helped.... values seem better... diff shows that the energy_collection in the cache is sorted inverse... might explain why the things went stray at 13:00 sunday, that was the first restart after installing .a2

but allas, initial is better... but issues came back

[ArnoutD]

Going back and forth to older version with or without removed cache doesn't solve it I will move my stick to my dev vm if I can replicate the issue there.

[brefra]

Looks like during a "rollover" of the current pulse counter into an energy log record, at least one of the energy log records is not used in the recalculation. Debug logging during a rollover should reveal what's actually going wrong....

[ArnoutD]

Tried to debug it yesterday.... saw some negative pulses but those dissappeared, moved the stick back to my prod env... issue gone.... got me baffled