rojamit

interesting that after exceptions like above python starts to crashing in other parts of code: ``` protocol: transport: Traceback (most recent call last): File "/usr/lib/python3.9/asyncio/selector_events.py", line 856, in _read_ready__data_received data...

``` Traceback (most recent call last): File "zvshield.py", line 598, in shared_dict_updater File "/usr/local/lib/python3.9/dist-packages/UltraDict/UltraDict.py", line 557, in __setitem__ self.apply_update() File "/usr/local/lib/python3.9/dist-packages/UltraDict/UltraDict.py", line 500, in apply_update self.load(force=True) File "/usr/local/lib/python3.9/dist-packages/UltraDict/UltraDict.py", line 450,...

reducing "changes" updates from "for change in changes: dict1['nested']['change'] = change" to "dict1['nested'] = changes" helped alot... not a fix but workaround

@ronny-rentner thanks, i'll test dev branch today my current buffer_size=500000 in my case (for now) there is 1 write/sec from master process, same for each child process, but child processes...

@ronny-rentner dev branch is much more stable. but... after about 12 hours, master has crashed: ``` future: Traceback (most recent call last): File "zvshield.py", line 515, in main File "zvshield.py",...

@ronny-rentner using dev branch, app is running fine for 2 days now (no restarts), definitely dev is much more stable: no errors from UltraDict during this period. I'm creating these...

@ronny-rentner yay just told all is ok and now after load spike app crashed again 40 processes, 200+k/s readings like: ``` sock, addr = server_socket.accept() if addr[0] in ADDR_INFO.data['banned']: ```...

@ronny-rentner i'll try print_status, but 'banned' is always-existing key... processes are crashing randomly on high load (not all in one time!), 'banned' is updating from master process using ADDR_INFO[banned] =...

@ronny-rentner also, recurse=True increased errors alot for now my best is: 1. Initializing like this: ``` ADDR_INFO = UltraDict({ 'banned': UltraDict({}, buffer_size=1000000), 'checked': {} }, buffer_size=1000000, recurse=False) ``` 2. Update...

@ronny-rentner yes, looks like error only appears with recurse=True (with dev branch) 'banned' key is always-existing key as i mentioned, it never gets deleted. In my case some processes crashes...