[CRASH] Executing mi("lb_reload") at timer_route

[apsaras]

Hello

I use timer_route to check every 10sec the status of a remote registrar server and when is down I execute queries to registrant and load_balancer DB Tables to re-register and re-route traffic to a secondary registrant server. When mi("reg_reload") is executed, it works. When mi("lb_reload") is executed service is crashing.

To reproduce.

1. Create a timer_route
2. Include mi("lb_reload");

Relevant System Logs

INFO:load_balancer:mi_lb_reload: "lb_reload" MI command received!
ERROR:core:db_use_table: invalid parameter value (nil), 0x7f460766d070
CRITICAL:core:sig_usr: segfault in process pid: 2900851, id: 8
INFO:core:handle_sigs: child process 2900851 exited by a signal 11
kernel: [8474889.624724] opensips[2900851]: segfault at 8 ip 00007f4607f06ab4 sp 00007ffc1bae4da8 error 6 in db_postgres.so[7f4607f04000+9000] likely on CPU 1 (core 0, socket 2)

OS/environment information

• Operating System: Debian 12
• OpenSIPS installation: packages
• other relevant information:

opensips -V
version: opensips 3.4.8 (x86_64/linux)
flags: STATS: On, DISABLE_NAGLE, USE_MCAST, SHM_MMAP, PKG_MALLOC, Q_MALLOC, F_MALLOC, HP_MALLOC, DBG_MALLOC, FAST_LOCK-ADAPTIVE_WAIT
ADAPTIVE_WAIT_LOOPS=1024, MAX_RECV_BUFFER_SIZE 262144, MAX_LISTEN 16, MAX_URI_SIZE 1024, BUF_SIZE 65535
poll method support: poll, epoll, sigio_rt, select.
git revision: 78cdea9
main.c compiled on with gcc 12

[bogdan-iancu]

@apsaras , this is (unfortunately) a known issue - the load_balancer modules does not need DB connections from the SIP worker processes at runtime, as it is not performing any runtime DB ops. All DB ops are done at startup or in the MI processes (for reload).
But with this new mi_script module, you may end up with having MI cmds called from a non MI processes, but from a regular SIP worker which has no DB con.
I recall @razvancrainea having a similar problem, I will check if there was any conclusion.