Memory leak when connecting to the LD relay proxy

[harshit98]

Is this a support request?
Yes

Describe the bug
When I connect my python service to LD relay proxy service - during pod starts few of the pods gets aggresively high memory utilisation reaching close to 100% and some other remaining pods face continuous increase in memory. The memory utilisation doesn't work in bounded manner. Without relay proxy, memory work fine.

To reproduce
Python code to connect relay proxy (using LD SDK Server pkg 7.4.1) and Python 3.10

def launch_darkly_setup():
    config_kwargs = dict(
        offline=Config.LAUNCHDARKLY_OFFLINE, sdk_key=Config.LAUNCHDARKLY_SDK_KEY
    )
    
    if Config.LAUNCHDARKLY_RELAY_ENABLED and Config.LAUNCHDARKLY_RELAY_URL:
        logging.info("LaunchDarkly Relay Proxy enabled, using relay at %s", Config.LAUNCHDARKLY_RELAY_URL)
        config_kwargs.update(
            stream_uri=Config.LAUNCHDARKLY_RELAY_URL,
            base_uri=Config.LAUNCHDARKLY_RELAY_URL,
            events_uri=Config.LAUNCHDARKLY_RELAY_URL,
            send_events=False,
        )
    else:
        logging.info("LaunchDarkly Relay Proxy disabled, connecting directly to LaunchDarkly")

    ldclient.set_config(ldclient.Config(**config_kwargs))

    # Wait for LaunchDarkly client to initialize (especially important for relay)
    client = ldclient.get()
    if not client.is_initialized():
        logger.info("Waiting for LaunchDarkly client to initialize...")
        client.wait_for_initialization(timeout=10)
        if client.is_initialized():
            logger.info("LaunchDarkly client initialized successfully")
        else:
            logger.warning("LaunchDarkly client initialization timeout - proceeding with defaults")
    else:
        logger.info("LaunchDarkly client already initialized")

Expected behavior
Service memory leak should not happen when connect LD client via relay proxy.

Logs
N/A

SDK version
7.6.1

Language version, developer tools
Python 3.10

OS/platform
N/A

Additional context
Profiler screenshots

Pod resources:

resources:
    limits:
        cpu: 1200m
        memory: 2Gi
    requests:
         cpu: 400m
         memory: 400Mi

[harshit98]

@keelerm84 Hi, while going through some closed issues in the python-server-sdk backlog, I noticed you've been quite active in the community. I'm running into a problem I haven't been able to resolve and I've tried several approaches within my understanding, but some aspects are beyond my current knowledge. Would you be open to helping me out?

[keelerm84]

@harshit98 yes I can definitely assist you with this. I won't be able to look at it until tomorrow unfortunately but I will let you know once I've learned more.

[harshit98]

@keelerm84 Thank you for acknowledging the issue. I'll wait for your reply.