### Description

After carrying out some load tests using the `Websocket` module, I noticed that sometimes the `websocket:closed` event is not triggered.

This problem seems to be particularly noticeable when the load is high and simultaneous. Example: 1000 records simultaneously.

In the tests carried out, I realized that the number of times the event is not triggered is small, in the order of [1-10]/1000.

I check that when I do a `ws.dump` I don't have any active connections, and I confirm that they are all effectively cleaned (`netstat -ano | grep 8061`). However, the event is not triggered for some connections. I can see this in the logs, and also when i display the content of an hash table, i'm dropping key inside `websocket:closed` event.

### Troubleshooting

The troubleshooting was simplistic, i.e. I ran the same test n times, and validated that the event was triggered for all connections, the number of `ws` connections listed by `Kamailio`, and the open connections.

#### Reproduction

To reproduce, we can use any mechanism that emulates the use of the `Websocket` module.

For example, I used the `sipexer` tool to create a `wss` connection with `Kamailio`, forcing simultaneous registrations.

Example command: `sipexer -cb -mt register -ex 60 -au <user> -ap <pass> -fuser example -fd mydomain -wso https://10.0.0.12:10000 -ruri sip:kamailio -su wss://kamailio:8061`

Kamailio TLS listener:

`listen=tls:10.0.0.12:8061 advertise PUBLIC_IP`

Websocket module:

``` loadmodule "websocket.so" tcp_accept_no_cl=yes ```

``` event_route[websocket:closed] { xlog("L_NOTICE", "WebSocket connection closed $proto:$si:$sp\n"); <I also delete where a htable key> } ```

#### Debugging Data

``` [root@ ~]$ prcmd ws.dump { connections: { } info: { wscounter: 0 truncated: no } } [root@ ~]$ netstat -ano | grep 8061 tcp 0 0 10.0.0.12:8061 0.0.0.0:* LISTEN off (0.00/0/0)

[root ~]$ prcmd htable.dump wsauth | grep "name" | wc -l 3

[root ~]$ cat /var/log/kamailio.log | grep "WebSocket connection closed" | wc -l 997

[root ~]$ cat /var/log/proxy-registrar/proxy-registrar.log | grep "ERR" | grep "WAR" | wc -l 0 ```

#### Log Messages

Nothing relevant.

#### SIP Traffic

Nothing relevant here, the SIP Flow is OK.

### Additional Information

``` version: kamailio 5.6.4 (x86_64/linux) a004cf-dirty flags: USE_TCP, USE_TLS, USE_SCTP, TLS_HOOKS, USE_RAW_SOCKS, DISABLE_NAGLE, USE_MCAST, DNS_IP_HACK, SHM_MMAP, PKG_MALLOC, Q_MALLOC, F_MALLOC, TLSF_MALLOC, DBG_SR_MEMORY, USE_FUTEX, FAST_LOCK-ADAPTIVE_WAIT, USE_DNS_CACHE, USE_DNS_FAILOVER, USE_NAPTR, USE_DST_BLOCKLIST, HAVE_RESOLV_RES ADAPTIVE_WAIT_LOOPS 1024, MAX_RECV_BUFFER_SIZE 262144, MAX_URI_SIZE 1024, BUF_SIZE 65535, DEFAULT PKG_SIZE 8MB poll method support: poll, epoll_lt, epoll_et, sigio_rt, select. id: a004cf -dirty compiled on 23:17:27 Jun 7 2024 with gcc 4.8.5 ```

* **Operating System**:

``` CentOS Linux release 7.9.2009 (Core) ```